Search is not available for this dataset
id
stringlengths 1
8
| text
stringlengths 72
9.81M
| addition_count
int64 0
10k
| commit_subject
stringlengths 0
3.7k
| deletion_count
int64 0
8.43k
| file_extension
stringlengths 0
32
| lang
stringlengths 1
94
| license
stringclasses 10
values | repo_name
stringlengths 9
59
|
|---|---|---|---|---|---|---|---|---|
10070150 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = '>= 2.2.2'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency "redis", ">= 4.2"
s.add_dependency "sinatra", ">= 1.2.6"
s.add_dependency "rubystats", ">= 0.3.0"
s.add_development_dependency "bundler", ">= 1.17"
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus
Only support Ruby versions >= 2.5
<DFF> @@ -22,7 +22,7 @@ Gem::Specification.new do |s|
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
- s.required_ruby_version = '>= 2.2.2'
+ s.required_ruby_version = '>= 2.5.0'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
| 1 | Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus | 1 | .gemspec | gemspec | mit | splitrb/split |
10070151 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = '>= 2.2.2'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency "redis", ">= 4.2"
s.add_dependency "sinatra", ">= 1.2.6"
s.add_dependency "rubystats", ">= 0.3.0"
s.add_development_dependency "bundler", ">= 1.17"
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus
Only support Ruby versions >= 2.5
<DFF> @@ -22,7 +22,7 @@ Gem::Specification.new do |s|
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
- s.required_ruby_version = '>= 2.2.2'
+ s.required_ruby_version = '>= 2.5.0'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
| 1 | Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus | 1 | .gemspec | gemspec | mit | splitrb/split |
10070152 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = '>= 2.2.2'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency "redis", ">= 4.2"
s.add_dependency "sinatra", ">= 1.2.6"
s.add_dependency "rubystats", ">= 0.3.0"
s.add_development_dependency "bundler", ">= 1.17"
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus
Only support Ruby versions >= 2.5
<DFF> @@ -22,7 +22,7 @@ Gem::Specification.new do |s|
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
- s.required_ruby_version = '>= 2.2.2'
+ s.required_ruby_version = '>= 2.5.0'
s.required_rubygems_version = '>= 2.0.0'
s.files = `git ls-files`.split("\n")
| 1 | Merge pull request #627 from splitrb/bump-ruby-support-to-2-5-plus | 1 | .gemspec | gemspec | mit | splitrb/split |
10070153 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_record_extra_info(metric_descriptor, key, value = 1)
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, _ = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments()
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Fix Style/DefWithParentheses
<DFF> @@ -105,7 +105,7 @@ module Split
Split.configuration.db_failover_on_db_error.call(e)
end
- def ab_active_experiments()
+ def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
| 1 | Fix Style/DefWithParentheses | 1 | .rb | rb | mit | splitrb/split |
10070154 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_record_extra_info(metric_descriptor, key, value = 1)
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, _ = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments()
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Fix Style/DefWithParentheses
<DFF> @@ -105,7 +105,7 @@ module Split
Split.configuration.db_failover_on_db_error.call(e)
end
- def ab_active_experiments()
+ def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
| 1 | Fix Style/DefWithParentheses | 1 | .rb | rb | mit | splitrb/split |
10070155 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_record_extra_info(metric_descriptor, key, value = 1)
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, _ = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments()
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Fix Style/DefWithParentheses
<DFF> @@ -105,7 +105,7 @@ module Split
Split.configuration.db_failover_on_db_error.call(e)
end
- def ab_active_experiments()
+ def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
| 1 | Fix Style/DefWithParentheses | 1 | .rb | rb | mit | splitrb/split |
10070156 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
### Rspec Helper
To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Recommended path for this file is 'spec/support/split_helper.rb'
module SplitHelper
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
```
Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Tweak RSpec instructions (#540)
<DFF> @@ -161,43 +161,39 @@ It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
-To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
-# Recommended path for this file is 'spec/support/split_helper.rb'
+# Create a file with these contents at 'spec/support/split_helper.rb'
+# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
+
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
- alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
-```
-Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
-```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
-Example:
+Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
-
use_ab_test experiment_name: "alternative_name"
-
post "/signups"
-
...
end
-
-# Force a specific experiment alternative to always be returned:
-# use_ab_test(signup_form: "single_page")
-#
-# Force alternatives for multiple experiments:
-# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
| 12 | Tweak RSpec instructions (#540) | 16 | .md | md | mit | splitrb/split |
10070157 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
### Rspec Helper
To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Recommended path for this file is 'spec/support/split_helper.rb'
module SplitHelper
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
```
Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Tweak RSpec instructions (#540)
<DFF> @@ -161,43 +161,39 @@ It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
-To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
-# Recommended path for this file is 'spec/support/split_helper.rb'
+# Create a file with these contents at 'spec/support/split_helper.rb'
+# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
+
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
- alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
-```
-Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
-```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
-Example:
+Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
-
use_ab_test experiment_name: "alternative_name"
-
post "/signups"
-
...
end
-
-# Force a specific experiment alternative to always be returned:
-# use_ab_test(signup_form: "single_page")
-#
-# Force alternatives for multiple experiments:
-# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
| 12 | Tweak RSpec instructions (#540) | 16 | .md | md | mit | splitrb/split |
10070158 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
### Rspec Helper
To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Recommended path for this file is 'spec/support/split_helper.rb'
module SplitHelper
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
```
Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Tweak RSpec instructions (#540)
<DFF> @@ -161,43 +161,39 @@ It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
-To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
-# Recommended path for this file is 'spec/support/split_helper.rb'
+# Create a file with these contents at 'spec/support/split_helper.rb'
+# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
+
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
- alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
end
end
end
-```
-Needs to ensure it `required` by rails_helper.rb or spec_helper.rb
-```ruby
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
-Example:
+Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
-
use_ab_test experiment_name: "alternative_name"
-
post "/signups"
-
...
end
-
-# Force a specific experiment alternative to always be returned:
-# use_ab_test(signup_form: "single_page")
-#
-# Force alternatives for multiple experiments:
-# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
```
| 12 | Tweak RSpec instructions (#540) | 16 | .md | md | mit | splitrb/split |
10070159 | <NME> split.rb
<BEF> require 'split/experiment'
require 'split/alternative'
require 'split/helper'
require 'split/version'
require 'split/configuration'
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
require "split/algorithms/whiplash"
require "split/alternative"
require "split/cache"
require "split/configuration"
require "split/encapsulated_helper"
require "split/exceptions"
require "split/experiment"
require "split/experiment_catalog"
require "split/extensions/string"
require "split/goals_collection"
require "split/helper"
require "split/combined_experiments_helper"
require "split/metric"
require "split/persistence"
require "split/redis_interface"
require "split/trial"
require "split/user"
require "split/version"
require "split/zscore"
require "split/engine" if defined?(Rails)
module Split
extend self
attr_accessor :configuration
# Accepts:
# 1. A redis URL (valid for `Redis.new(url: url)`)
# 2. an options hash compatible with `Redis.new`
# 3. or a valid Redis instance (one that responds to `#smembers`). Likely,
# this will be an instance of either `Redis`, `Redis::Client`,
# `Redis::DistRedis`, or `Redis::Namespace`.
def redis=(server)
@redis = if server.is_a?(String)
Redis.new(url: server)
elsif server.is_a?(Hash)
Redis.new(server)
elsif server.respond_to?(:smembers)
server
else
raise ArgumentError,
"You must supply a url, options hash or valid Redis connection instance"
end
end
# Returns the current Redis connection. If none has been created, will
# create a new one.
def redis
return @redis if @redis
self.redis = self.configuration.redis
self.redis
end
# Call this method to modify defaults in your initializers.
#
# @example
# Split.configure do |config|
# config.ignore_ip_addresses = '192.168.2.1'
# end
def configure
self.configuration ||= Configuration.new
yield(configuration)
end
def cache(namespace, key, &block)
Split::Cache.fetch(namespace, key, &block)
end
end
# Check to see if being run in a Rails application. If so, wait until before_initialize to run configuration so Gems that create ENV variables have the chance to initialize first.
if defined?(::Rails)
class Split::Railtie < Rails::Railtie
config.before_initialize { Split.configure { } }
end
else
Split.configure { }
end
<MSG> rolled require statements into iterator
<DFF> @@ -1,8 +1,7 @@
-require 'split/experiment'
-require 'split/alternative'
-require 'split/helper'
-require 'split/version'
-require 'split/configuration'
+%w[experiment alternative helper version configuration].each do |f|
+ require "split/#{f}"
+end
+
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
| 4 | rolled require statements into iterator | 5 | .rb | rb | mit | splitrb/split |
10070160 | <NME> split.rb
<BEF> require 'split/experiment'
require 'split/alternative'
require 'split/helper'
require 'split/version'
require 'split/configuration'
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
require "split/algorithms/whiplash"
require "split/alternative"
require "split/cache"
require "split/configuration"
require "split/encapsulated_helper"
require "split/exceptions"
require "split/experiment"
require "split/experiment_catalog"
require "split/extensions/string"
require "split/goals_collection"
require "split/helper"
require "split/combined_experiments_helper"
require "split/metric"
require "split/persistence"
require "split/redis_interface"
require "split/trial"
require "split/user"
require "split/version"
require "split/zscore"
require "split/engine" if defined?(Rails)
module Split
extend self
attr_accessor :configuration
# Accepts:
# 1. A redis URL (valid for `Redis.new(url: url)`)
# 2. an options hash compatible with `Redis.new`
# 3. or a valid Redis instance (one that responds to `#smembers`). Likely,
# this will be an instance of either `Redis`, `Redis::Client`,
# `Redis::DistRedis`, or `Redis::Namespace`.
def redis=(server)
@redis = if server.is_a?(String)
Redis.new(url: server)
elsif server.is_a?(Hash)
Redis.new(server)
elsif server.respond_to?(:smembers)
server
else
raise ArgumentError,
"You must supply a url, options hash or valid Redis connection instance"
end
end
# Returns the current Redis connection. If none has been created, will
# create a new one.
def redis
return @redis if @redis
self.redis = self.configuration.redis
self.redis
end
# Call this method to modify defaults in your initializers.
#
# @example
# Split.configure do |config|
# config.ignore_ip_addresses = '192.168.2.1'
# end
def configure
self.configuration ||= Configuration.new
yield(configuration)
end
def cache(namespace, key, &block)
Split::Cache.fetch(namespace, key, &block)
end
end
# Check to see if being run in a Rails application. If so, wait until before_initialize to run configuration so Gems that create ENV variables have the chance to initialize first.
if defined?(::Rails)
class Split::Railtie < Rails::Railtie
config.before_initialize { Split.configure { } }
end
else
Split.configure { }
end
<MSG> rolled require statements into iterator
<DFF> @@ -1,8 +1,7 @@
-require 'split/experiment'
-require 'split/alternative'
-require 'split/helper'
-require 'split/version'
-require 'split/configuration'
+%w[experiment alternative helper version configuration].each do |f|
+ require "split/#{f}"
+end
+
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
| 4 | rolled require statements into iterator | 5 | .rb | rb | mit | splitrb/split |
10070161 | <NME> split.rb
<BEF> require 'split/experiment'
require 'split/alternative'
require 'split/helper'
require 'split/version'
require 'split/configuration'
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
require "split/algorithms/whiplash"
require "split/alternative"
require "split/cache"
require "split/configuration"
require "split/encapsulated_helper"
require "split/exceptions"
require "split/experiment"
require "split/experiment_catalog"
require "split/extensions/string"
require "split/goals_collection"
require "split/helper"
require "split/combined_experiments_helper"
require "split/metric"
require "split/persistence"
require "split/redis_interface"
require "split/trial"
require "split/user"
require "split/version"
require "split/zscore"
require "split/engine" if defined?(Rails)
module Split
extend self
attr_accessor :configuration
# Accepts:
# 1. A redis URL (valid for `Redis.new(url: url)`)
# 2. an options hash compatible with `Redis.new`
# 3. or a valid Redis instance (one that responds to `#smembers`). Likely,
# this will be an instance of either `Redis`, `Redis::Client`,
# `Redis::DistRedis`, or `Redis::Namespace`.
def redis=(server)
@redis = if server.is_a?(String)
Redis.new(url: server)
elsif server.is_a?(Hash)
Redis.new(server)
elsif server.respond_to?(:smembers)
server
else
raise ArgumentError,
"You must supply a url, options hash or valid Redis connection instance"
end
end
# Returns the current Redis connection. If none has been created, will
# create a new one.
def redis
return @redis if @redis
self.redis = self.configuration.redis
self.redis
end
# Call this method to modify defaults in your initializers.
#
# @example
# Split.configure do |config|
# config.ignore_ip_addresses = '192.168.2.1'
# end
def configure
self.configuration ||= Configuration.new
yield(configuration)
end
def cache(namespace, key, &block)
Split::Cache.fetch(namespace, key, &block)
end
end
# Check to see if being run in a Rails application. If so, wait until before_initialize to run configuration so Gems that create ENV variables have the chance to initialize first.
if defined?(::Rails)
class Split::Railtie < Rails::Railtie
config.before_initialize { Split.configure { } }
end
else
Split.configure { }
end
<MSG> rolled require statements into iterator
<DFF> @@ -1,8 +1,7 @@
-require 'split/experiment'
-require 'split/alternative'
-require 'split/helper'
-require 'split/version'
-require 'split/configuration'
+%w[experiment alternative helper version configuration].each do |f|
+ require "split/#{f}"
+end
+
require 'split/engine' if defined?(Rails)
require 'redis/namespace'
| 4 | rolled require statements into iterator | 5 | .rb | rb | mit | splitrb/split |
10070162 | <NME> version.rb
<BEF> module Split
MAJOR = 0
MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> Version 0.6.1
<DFF> @@ -1,6 +1,6 @@
module Split
MAJOR = 0
MINOR = 6
- PATCH = 0
+ PATCH = 1
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | Version 0.6.1 | 1 | .rb | rb | mit | splitrb/split |
10070163 | <NME> version.rb
<BEF> module Split
MAJOR = 0
MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> Version 0.6.1
<DFF> @@ -1,6 +1,6 @@
module Split
MAJOR = 0
MINOR = 6
- PATCH = 0
+ PATCH = 1
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | Version 0.6.1 | 1 | .rb | rb | mit | splitrb/split |
10070164 | <NME> version.rb
<BEF> module Split
MAJOR = 0
MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> Version 0.6.1
<DFF> @@ -1,6 +1,6 @@
module Split
MAJOR = 0
MINOR = 6
- PATCH = 0
+ PATCH = 1
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | Version 0.6.1 | 1 | .rb | rb | mit | splitrb/split |
10070165 | <NME> buildout.cfg
<BEF> [buildout]
parts = django
find-links = http://bitbucket.org/ubernostrum/django-registration/downloads/django-registration-0.8-alpha-1.tar.gz
unzip = true
eggs = pkginfo
django-registration==0.8-alpha-1
[django]
recipe = djangorecipe
version = 1.1.1
settings = development
eggs = ${buildout:eggs}
test = djangopypi
project = chishop
wsgi = true
<MSG> Made settings.py default
<DFF> @@ -8,7 +8,7 @@ eggs = pkginfo
[django]
recipe = djangorecipe
version = 1.1.1
-settings = development
+settings = settings
eggs = ${buildout:eggs}
test = djangopypi
project = chishop
| 1 | Made settings.py default | 1 | .cfg | cfg | bsd-3-clause | ask/chishop |
10070166 | <NME> experiment.rb
<BEF> module Split
class Experiment
attr_accessor :name
attr_accessor :alternatives
attr_accessor :winner
def initialize(name, *alternatives)
@name = name.to_s
@alternatives = alternatives
end
def winner
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def alternatives
@alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
def save
Split.redis.sadd(:experiments, name)
@alternatives.each {|a| Split.redis.sadd(name, a) }
end
def self.all
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> renamed alternatives accessor on Experiment to alternative_names
<DFF> @@ -1,12 +1,12 @@
module Split
class Experiment
attr_accessor :name
- attr_accessor :alternatives
+ attr_accessor :alternative_names
attr_accessor :winner
- def initialize(name, *alternatives)
+ def initialize(name, *alternative_names)
@name = name.to_s
- @alternatives = alternatives
+ @alternative_names = alternative_names
end
def winner
@@ -30,7 +30,7 @@ module Split
end
def alternatives
- @alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
+ @alternative_names.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
@@ -46,7 +46,7 @@ module Split
def save
Split.redis.sadd(:experiments, name)
- @alternatives.each {|a| Split.redis.sadd(name, a) }
+ @alternative_names.each {|a| Split.redis.sadd(name, a) }
end
def self.all
| 5 | renamed alternatives accessor on Experiment to alternative_names | 5 | .rb | rb | mit | splitrb/split |
10070167 | <NME> experiment.rb
<BEF> module Split
class Experiment
attr_accessor :name
attr_accessor :alternatives
attr_accessor :winner
def initialize(name, *alternatives)
@name = name.to_s
@alternatives = alternatives
end
def winner
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def alternatives
@alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
def save
Split.redis.sadd(:experiments, name)
@alternatives.each {|a| Split.redis.sadd(name, a) }
end
def self.all
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> renamed alternatives accessor on Experiment to alternative_names
<DFF> @@ -1,12 +1,12 @@
module Split
class Experiment
attr_accessor :name
- attr_accessor :alternatives
+ attr_accessor :alternative_names
attr_accessor :winner
- def initialize(name, *alternatives)
+ def initialize(name, *alternative_names)
@name = name.to_s
- @alternatives = alternatives
+ @alternative_names = alternative_names
end
def winner
@@ -30,7 +30,7 @@ module Split
end
def alternatives
- @alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
+ @alternative_names.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
@@ -46,7 +46,7 @@ module Split
def save
Split.redis.sadd(:experiments, name)
- @alternatives.each {|a| Split.redis.sadd(name, a) }
+ @alternative_names.each {|a| Split.redis.sadd(name, a) }
end
def self.all
| 5 | renamed alternatives accessor on Experiment to alternative_names | 5 | .rb | rb | mit | splitrb/split |
10070168 | <NME> experiment.rb
<BEF> module Split
class Experiment
attr_accessor :name
attr_accessor :alternatives
attr_accessor :winner
def initialize(name, *alternatives)
@name = name.to_s
@alternatives = alternatives
end
def winner
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def alternatives
@alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
def save
Split.redis.sadd(:experiments, name)
@alternatives.each {|a| Split.redis.sadd(name, a) }
end
def self.all
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> renamed alternatives accessor on Experiment to alternative_names
<DFF> @@ -1,12 +1,12 @@
module Split
class Experiment
attr_accessor :name
- attr_accessor :alternatives
+ attr_accessor :alternative_names
attr_accessor :winner
- def initialize(name, *alternatives)
+ def initialize(name, *alternative_names)
@name = name.to_s
- @alternatives = alternatives
+ @alternative_names = alternative_names
end
def winner
@@ -30,7 +30,7 @@ module Split
end
def alternatives
- @alternatives.map {|a| Split::Alternative.find_or_create(a, name)}
+ @alternative_names.map {|a| Split::Alternative.find_or_create(a, name)}
end
def next_alternative
@@ -46,7 +46,7 @@ module Split
def save
Split.redis.sadd(:experiments, name)
- @alternatives.each {|a| Split.redis.sadd(name, a) }
+ @alternative_names.each {|a| Split.redis.sadd(name, a) }
end
def self.all
| 5 | renamed alternatives accessor on Experiment to alternative_names | 5 | .rb | rb | mit | splitrb/split |
10070169 | <NME> paginator_spec.rb
<BEF> ADDFILE
<MSG> dashboard/paginator (#518)
<DFF> @@ -0,0 +1,37 @@
+# frozen_string_literal: true
+require 'spec_helper'
+require 'split/dashboard/paginator'
+
+describe Split::DashboardPaginator do
+ context 'when collection is 1..20' do
+ let(:collection) { (1..20).to_a }
+
+ context 'when per 5 for page' do
+ let(:per) { 5 }
+
+ it 'when page number is 1 result is [1, 2, 3, 4, 5]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5]
+ end
+
+ it 'when page number is 2 result is [6, 7, 8, 9, 10]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [6, 7, 8, 9, 10]
+ end
+ end
+
+ context 'when per 10 for page' do
+ let(:per) { 10 }
+
+ it 'when page number is 1 result is [1..10]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
+ end
+
+ it 'when page number is 2 result is [10..20]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
+ end
+ end
+ end
+end
| 37 | dashboard/paginator (#518) | 0 | .rb | rb | mit | splitrb/split |
10070170 | <NME> paginator_spec.rb
<BEF> ADDFILE
<MSG> dashboard/paginator (#518)
<DFF> @@ -0,0 +1,37 @@
+# frozen_string_literal: true
+require 'spec_helper'
+require 'split/dashboard/paginator'
+
+describe Split::DashboardPaginator do
+ context 'when collection is 1..20' do
+ let(:collection) { (1..20).to_a }
+
+ context 'when per 5 for page' do
+ let(:per) { 5 }
+
+ it 'when page number is 1 result is [1, 2, 3, 4, 5]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5]
+ end
+
+ it 'when page number is 2 result is [6, 7, 8, 9, 10]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [6, 7, 8, 9, 10]
+ end
+ end
+
+ context 'when per 10 for page' do
+ let(:per) { 10 }
+
+ it 'when page number is 1 result is [1..10]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
+ end
+
+ it 'when page number is 2 result is [10..20]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
+ end
+ end
+ end
+end
| 37 | dashboard/paginator (#518) | 0 | .rb | rb | mit | splitrb/split |
10070171 | <NME> paginator_spec.rb
<BEF> ADDFILE
<MSG> dashboard/paginator (#518)
<DFF> @@ -0,0 +1,37 @@
+# frozen_string_literal: true
+require 'spec_helper'
+require 'split/dashboard/paginator'
+
+describe Split::DashboardPaginator do
+ context 'when collection is 1..20' do
+ let(:collection) { (1..20).to_a }
+
+ context 'when per 5 for page' do
+ let(:per) { 5 }
+
+ it 'when page number is 1 result is [1, 2, 3, 4, 5]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5]
+ end
+
+ it 'when page number is 2 result is [6, 7, 8, 9, 10]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [6, 7, 8, 9, 10]
+ end
+ end
+
+ context 'when per 10 for page' do
+ let(:per) { 10 }
+
+ it 'when page number is 1 result is [1..10]' do
+ result = Split::DashboardPaginator.new(collection, 1, per).paginate
+ expect(result).to eql [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
+ end
+
+ it 'when page number is 2 result is [10..20]' do
+ result = Split::DashboardPaginator.new(collection, 2, per).paginate
+ expect(result).to eql [11, 12, 13, 14, 15, 16, 17, 18, 19, 20]
+ end
+ end
+ end
+end
| 37 | dashboard/paginator (#518) | 0 | .rb | rb | mit | splitrb/split |
10070172 | <NME> transducers.js
<BEF>
// basic protocol helpers
var symbolExists = typeof Symbol !== 'undefined';
var protocols = {
iterator: symbolExists ? Symbol.iterator : '@@iterator'
};
function throwProtocolError(name, coll) {
throw new Error("don't know how to " + name + " collection: " +
coll);
}
function fulfillsProtocol(obj, name) {
if(name === 'iterator') {
// Accept ill-formed iterators that don't conform to the
// protocol by accepting just next()
return obj[protocols.iterator] || obj.next;
}
return obj[protocols[name]];
}
function getProtocolProperty(obj, name) {
return obj[protocols[name]];
}
function iterator(coll) {
var iter = getProtocolProperty(coll, 'iterator');
if(iter) {
return iter.call(coll);
}
else if(coll.next) {
// Basic duck typing to accept an ill-formed iterator that doesn't
// conform to the iterator protocol (all iterators should have the
// @@iterator method and return themselves, but some engines don't
// have that on generators like older v8)
return coll;
}
else if(isArray(coll)) {
return new ArrayIterator(coll);
}
else if(isObject(coll)) {
return new ObjectIterator(coll);
}
}
function ArrayIterator(arr) {
this.arr = arr;
this.index = 0;
}
ArrayIterator.prototype.next = function() {
if(this.index < this.arr.length) {
return {
value: this.arr[this.index++],
done: false
};
}
return {
done: true
}
};
function ObjectIterator(obj) {
this.obj = obj;
this.keys = Object.keys(obj);
this.index = 0;
}
ObjectIterator.prototype.next = function() {
if(this.index < this.keys.length) {
var k = this.keys[this.index++];
return {
value: [k, this.obj[k]],
done: false
};
}
return {
done: true
}
};
// helpers
var toString = Object.prototype.toString;
var isArray = typeof Array.isArray === 'function' ? Array.isArray : function(obj) {
return toString.call(obj) == '[object Array]';
};
function isFunction(x) {
return typeof x === 'function';
}
function isObject(x) {
return x instanceof Object &&
Object.getPrototypeOf(x) === Object.getPrototypeOf({});
}
function isNumber(x) {
return typeof x === 'number';
}
function Reduced(value) {
this['@@transducer/reduced'] = true;
this['@@transducer/value'] = value;
}
function isReduced(x) {
return (x instanceof Reduced) || (x && x['@@transducer/reduced']);
}
function deref(x) {
while(++index < len) {
result = xform.step(result, coll[index]);
if(result instanceof Reduced) {
return result.val;
}
}
return xform.result(result);
if(isReduced(val)) {
return val;
} else {
return new Reduced(val);
}
while(!val.done) {
result = xform.step(result, val.value);
if(result instanceof Reduced) {
return result.val;
}
val = iter.next();
}
if(isReduced(v)) {
return deref(v);
} else {
return v;
}
}
function reduce(coll, xform, init) {
if(isArray(coll)) {
var result = init;
var index = -1;
var len = coll.length;
while(++index < len) {
result = xform['@@transducer/step'](result, coll[index]);
if(isReduced(result)) {
result = deref(result);
break;
}
}
return xform['@@transducer/result'](result);
}
else if(isObject(coll) || fulfillsProtocol(coll, 'iterator')) {
var result = init;
var iter = iterator(coll);
var val = iter.next();
while(!val.done) {
result = xform['@@transducer/step'](result, val.value);
if(isReduced(result)) {
result = deref(result);
break;
}
val = iter.next();
}
return xform['@@transducer/result'](result);
}
throwProtocolError('iterate', coll);
}
function transduce(coll, xform, reducer, init) {
xform = xform(reducer);
if(init === undefined) {
init = xform['@@transducer/init']();
}
return reduce(coll, xform, init);
}
function compose() {
var funcs = Array.prototype.slice.call(arguments);
return function(r) {
var value = r;
for(var i=funcs.length-1; i>=0; i--) {
value = funcs[i](value);
}
return value;
}
}
// transformations
function transformer(f) {
var t = {};
t['@@transducer/init'] = function() {
throw new Error('init value unavailable');
};
t['@@transducer/result'] = function(v) {
return v;
};
t['@@transducer/step'] = f;
return t;
}
function bound(f, ctx, count) {
count = count != null ? count : 1;
if(!ctx) {
return f;
}
else {
switch(count) {
case 1:
return function(x) {
return f.call(ctx, x);
}
case 2:
return function(x, y) {
return f.call(ctx, x, y);
}
default:
return f.bind(ctx);
}
}
}
function arrayMap(arr, f, ctx) {
var index = -1;
var length = arr.length;
var result = Array(length);
f = bound(f, ctx, 2);
while (++index < length) {
result[index] = f(arr[index], index);
}
return result;
}
function arrayFilter(arr, f, ctx) {
var len = arr.length;
var result = [];
f = bound(f, ctx, 2);
for(var i=0; i<len; i++) {
if(f(arr[i], i)) {
result.push(arr[i]);
}
}
return result;
}
function Map(f, xform) {
this.xform = xform;
this.f = f;
}
Map.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Map.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Map.prototype['@@transducer/step'] = function(res, input) {
return this.xform['@@transducer/step'](res, this.f(input));
};
function map(coll, f, ctx) {
if(isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
if(coll) {
if(isArray(coll)) {
return arrayMap(coll, f, ctx);
}
return seq(coll, map(f));
}
return function(xform) {
return new Map(f, xform);
}
}
function Filter(f, xform) {
this.xform = xform;
this.f = f;
}
Filter.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Filter.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Filter.prototype['@@transducer/step'] = function(res, input) {
if(this.f(input)) {
return this.xform['@@transducer/step'](res, input);
}
return res;
};
function filter(coll, f, ctx) {
if(isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
if(coll) {
if(isArray(coll)) {
return arrayFilter(coll, f, ctx);
}
return seq(coll, filter(f));
}
return function(xform) {
return new Filter(f, xform);
};
}
function remove(coll, f, ctx) {
if(isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
return filter(coll, function(x) { return !f(x); });
}
function keep(coll) {
return filter(coll, function(x) { return x != null });
}
function Dedupe(xform) {
this.xform = xform;
this.last = undefined;
}
Dedupe.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Dedupe.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Dedupe.prototype['@@transducer/step'] = function(result, input) {
if(input !== this.last) {
this.last = input;
return this.xform['@@transducer/step'](result, input);
}
return result;
};
function dedupe(coll) {
if(coll) {
return seq(coll, dedupe());
}
return function(xform) {
return new Dedupe(xform);
}
}
function TakeWhile(f, xform) {
this.xform = xform;
this.f = f;
}
TakeWhile.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
TakeWhile.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
TakeWhile.prototype['@@transducer/step'] = function(result, input) {
if(this.f(input)) {
return this.xform['@@transducer/step'](result, input);
}
return new Reduced(result);
};
function takeWhile(coll, f, ctx) {
if(isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
if(coll) {
return seq(coll, takeWhile(f));
}
return function(xform) {
return new TakeWhile(f, xform);
}
}
function Take(n, xform) {
this.n = n;
this.i = 0;
this.xform = xform;
}
Take.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Take.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Take.prototype['@@transducer/step'] = function(result, input) {
if (this.i < this.n) {
result = this.xform['@@transducer/step'](result, input);
if(this.i + 1 >= this.n) {
// Finish reducing on the same step as the final value. TODO:
// double-check that this doesn't break any semantics
result = ensureReduced(result);
}
}
this.i++;
return result;
};
function take(coll, n) {
if(isNumber(coll)) { n = coll; coll = null }
if(coll) {
return seq(coll, take(n));
}
return function(xform) {
return new Take(n, xform);
}
}
function Drop(n, xform) {
this.n = n;
this.i = 0;
this.xform = xform;
}
Drop.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Drop.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Drop.prototype['@@transducer/step'] = function(result, input) {
if(this.i++ < this.n) {
return result;
}
return this.xform['@@transducer/step'](result, input);
};
function drop(coll, n) {
if(isNumber(coll)) { n = coll; coll = null }
if(coll) {
return seq(coll, drop(n));
}
return function(xform) {
return new Drop(n, xform);
}
}
function DropWhile(f, xform) {
this.xform = xform;
this.f = f;
this.dropping = true;
}
DropWhile.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
DropWhile.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
DropWhile.prototype['@@transducer/step'] = function(result, input) {
if(this.dropping) {
if(this.f(input)) {
return result;
}
else {
this.dropping = false;
}
}
return this.xform['@@transducer/step'](result, input);
};
function dropWhile(coll, f, ctx) {
if(isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
if(coll) {
return seq(coll, dropWhile(f));
}
return function(xform) {
return new DropWhile(f, xform);
}
}
function Partition(n, xform) {
this.n = n;
this.i = 0;
this.xform = xform;
this.part = new Array(n);
}
Partition.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Partition.prototype['@@transducer/result'] = function(v) {
if (this.i > 0) {
return ensureUnreduced(this.xform['@@transducer/step'](v, this.part.slice(0, this.i)));
}
return this.xform['@@transducer/result'](v);
};
Partition.prototype['@@transducer/step'] = function(result, input) {
this.part[this.i] = input;
this.i += 1;
if (this.i === this.n) {
var out = this.part.slice(0, this.n);
this.part = new Array(this.n);
this.i = 0;
return this.xform['@@transducer/step'](result, out);
}
return result;
};
function partition(coll, n) {
if (isNumber(coll)) {
n = coll; coll = null;
}
if (coll) {
return seq(coll, partition(n));
}
return function(xform) {
return new Partition(n, xform);
};
}
var NOTHING = {};
function PartitionBy(f, xform) {
// TODO: take an "opts" object that allows the user to specify
// equality
this.f = f;
this.xform = xform;
this.part = [];
this.last = NOTHING;
}
PartitionBy.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
PartitionBy.prototype['@@transducer/result'] = function(v) {
var l = this.part.length;
if (l > 0) {
return ensureUnreduced(this.xform['@@transducer/step'](v, this.part.slice(0, l)));
}
return this.xform['@@transducer/result'](v);
};
PartitionBy.prototype['@@transducer/step'] = function(result, input) {
var current = this.f(input);
if (current === this.last || this.last === NOTHING) {
this.part.push(input);
} else {
result = this.xform['@@transducer/step'](result, this.part);
this.part = [input];
}
this.last = current;
return result;
};
function partitionBy(coll, f, ctx) {
if (isFunction(coll)) { ctx = f; f = coll; coll = null; }
f = bound(f, ctx);
if (coll) {
return seq(coll, partitionBy(f));
}
return function(xform) {
return new PartitionBy(f, xform);
};
}
function Interpose(sep, xform) {
this.sep = sep;
this.xform = xform;
this.started = false;
}
Interpose.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Interpose.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Interpose.prototype['@@transducer/step'] = function(result, input) {
if (this.started) {
var withSep = this.xform['@@transducer/step'](result, this.sep);
if (isReduced(withSep)) {
return withSep;
} else {
return this.xform['@@transducer/step'](withSep, input);
}
} else {
this.started = true;
return this.xform['@@transducer/step'](result, input);
}
};
/**
* Returns a new collection containing elements of the given
* collection, separated by the specified separator. Returns a
* transducer if a collection is not provided.
*/
function interpose(coll, separator) {
if (arguments.length === 1) {
separator = coll;
return function(xform) {
return new Interpose(separator, xform);
};
}
return seq(coll, interpose(separator));
}
function Repeat(n, xform) {
this.xform = xform;
this.n = n;
}
Repeat.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Repeat.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Repeat.prototype['@@transducer/step'] = function(result, input) {
var n = this.n;
var r = result;
for (var i = 0; i < n; i++) {
r = this.xform['@@transducer/step'](r, input);
if (isReduced(r)) {
break;
}
}
return r;
};
/**
* Returns a new collection containing elements of the given
* collection, each repeated n times. Returns a transducer if a
* collection is not provided.
*/
function repeat(coll, n) {
if (arguments.length === 1) {
n = coll;
return function(xform) {
return new Repeat(n, xform);
};
}
return seq(coll, repeat(n));
}
function TakeNth(n, xform) {
this.xform = xform;
this.n = n;
this.i = -1;
}
TakeNth.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
TakeNth.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
TakeNth.prototype['@@transducer/step'] = function(result, input) {
this.i += 1;
if (this.i % this.n === 0) {
return this.xform['@@transducer/step'](result, input);
}
return result;
};
/**
* Returns a new collection of every nth element of the given
* collection. Returns a transducer if a collection is not provided.
*/
function takeNth(coll, nth) {
if (arguments.length === 1) {
nth = coll;
return function(xform) {
return new TakeNth(nth, xform);
};
}
return seq(coll, takeNth(nth));
}
// pure transducers (cannot take collections)
function Cat(xform) {
this.xform = xform;
}
Cat.prototype['@@transducer/init'] = function() {
return this.xform['@@transducer/init']();
};
Cat.prototype['@@transducer/result'] = function(v) {
return this.xform['@@transducer/result'](v);
};
Cat.prototype['@@transducer/step'] = function(result, input) {
var xform = this.xform;
var newxform = {};
newxform['@@transducer/init'] = function() {
return xform['@@transducer/init']();
};
newxform['@@transducer/result'] = function(v) {
return v;
};
newxform['@@transducer/step'] = function(result, input) {
var val = xform['@@transducer/step'](result, input);
return isReduced(val) ? deref(val) : val;
};
return reduce(input, newxform, result);
};
function cat(xform) {
return new Cat(xform);
}
function mapcat(f, ctx) {
f = bound(f, ctx);
return compose(map(f), cat);
}
// collection helpers
function push(arr, x) {
arr.push(x);
return arr;
}
function merge(obj, x) {
if(isArray(x) && x.length === 2) {
obj[x[0]] = x[1];
}
else {
var keys = Object.keys(x);
var len = keys.length;
for(var i=0; i<len; i++) {
obj[keys[i]] = x[keys[i]];
}
}
return obj;
}
var arrayReducer = {};
arrayReducer['@@transducer/init'] = function() {
return [];
};
arrayReducer['@@transducer/result'] = function(v) {
return v;
};
arrayReducer['@@transducer/step'] = push;
var objReducer = {};
objReducer['@@transducer/init'] = function() {
return {};
};
objReducer['@@transducer/result'] = function(v) {
return v;
};
objReducer['@@transducer/step'] = merge;
// building new collections
function toArray(coll, xform) {
if(!xform) {
return reduce(coll, arrayReducer, []);
}
return transduce(coll, xform, arrayReducer, []);
}
function toObj(coll, xform) {
if(!xform) {
return reduce(coll, objReducer, {});
}
return transduce(coll, xform, objReducer, {});
}
function toIter(coll, xform) {
if(!xform) {
return iterator(coll);
}
return new LazyTransformer(xform, coll);
}
function seq(coll, xform) {
if(isArray(coll)) {
return transduce(coll, xform, arrayReducer, []);
}
else if(isObject(coll)) {
return transduce(coll, xform, objReducer, {});
}
else if(coll['@@transducer/step']) {
var init;
if(coll['@@transducer/init']) {
init = coll['@@transducer/init']();
}
else {
init = new coll.constructor();
}
return transduce(coll, xform, coll, init);
}
else if(fulfillsProtocol(coll, 'iterator')) {
return new LazyTransformer(xform, coll);
}
throwProtocolError('sequence', coll);
}
function into(to, xform, from) {
if(isArray(to)) {
return transduce(from, xform, arrayReducer, to);
}
else if(isObject(to)) {
return transduce(from, xform, objReducer, to);
}
else if(to['@@transducer/step']) {
return transduce(from,
xform,
to,
to);
}
throwProtocolError('into', to);
}
// laziness
var stepper = {};
stepper['@@transducer/result'] = function(v) {
return isReduced(v) ? deref(v) : v;
};
stepper['@@transducer/step'] = function(lt, x) {
lt.items.push(x);
return lt.rest;
};
function Stepper(xform, iter) {
this.xform = xform(stepper);
this.iter = iter;
}
Stepper.prototype['@@transducer/step'] = function(lt) {
var len = lt.items.length;
while(lt.items.length === len) {
var n = this.iter.next();
if(n.done || isReduced(n.value)) {
// finalize
this.xform['@@transducer/result'](this);
break;
}
// step
this.xform['@@transducer/step'](lt, n.value);
}
}
function LazyTransformer(xform, coll) {
this.iter = iterator(coll);
this.items = [];
this.stepper = new Stepper(xform, iterator(coll));
}
LazyTransformer.prototype[protocols.iterator] = function() {
return this;
}
LazyTransformer.prototype.next = function() {
this['@@transducer/step']();
if(this.items.length) {
return {
value: this.items.pop(),
done: false
}
}
else {
return { done: true };
}
};
LazyTransformer.prototype['@@transducer/step'] = function() {
if(!this.items.length) {
this.stepper['@@transducer/step'](this);
}
}
// util
function range(n) {
var arr = new Array(n);
for(var i=0; i<arr.length; i++) {
arr[i] = i;
}
return arr;
}
module.exports = {
reduce: reduce,
transformer: transformer,
Reduced: Reduced,
isReduced: isReduced,
iterator: iterator,
push: push,
merge: merge,
transduce: transduce,
seq: seq,
toArray: toArray,
toObj: toObj,
toIter: toIter,
into: into,
compose: compose,
map: map,
filter: filter,
remove: remove,
cat: cat,
mapcat: mapcat,
keep: keep,
dedupe: dedupe,
take: take,
takeWhile: takeWhile,
takeNth: takeNth,
drop: drop,
dropWhile: dropWhile,
partition: partition,
partitionBy: partitionBy,
interpose: interpose,
repeat: repeat,
range: range,
LazyTransformer: LazyTransformer
};
<MSG> Merge pull request #15 from kevinbeaty/master
Ensure xform.result called on Reduced.val
<DFF> @@ -115,7 +115,8 @@ function reduce(coll, xform, init) {
while(++index < len) {
result = xform.step(result, coll[index]);
if(result instanceof Reduced) {
- return result.val;
+ result = result.val;
+ break;
}
}
return xform.result(result);
@@ -127,7 +128,8 @@ function reduce(coll, xform, init) {
while(!val.done) {
result = xform.step(result, val.value);
if(result instanceof Reduced) {
- return result.val;
+ result = result.val;
+ break;
}
val = iter.next();
}
| 4 | Merge pull request #15 from kevinbeaty/master | 2 | .js | js | bsd-2-clause | jlongster/transducers.js |
10070173 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
describe Multivariate::Helper do
before(:each) do
REDIS.flushall
Multivariate::Helper.current_user = {}
end
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Improved redis configuration by copying from resque
<DFF> @@ -2,7 +2,7 @@ require 'spec_helper'
describe Multivariate::Helper do
before(:each) do
- REDIS.flushall
+ Multivariate.redis.flushall
Multivariate::Helper.current_user = {}
end
| 1 | Improved redis configuration by copying from resque | 1 | .rb | rb | mit | splitrb/split |
10070174 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
describe Multivariate::Helper do
before(:each) do
REDIS.flushall
Multivariate::Helper.current_user = {}
end
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Improved redis configuration by copying from resque
<DFF> @@ -2,7 +2,7 @@ require 'spec_helper'
describe Multivariate::Helper do
before(:each) do
- REDIS.flushall
+ Multivariate.redis.flushall
Multivariate::Helper.current_user = {}
end
| 1 | Improved redis configuration by copying from resque | 1 | .rb | rb | mit | splitrb/split |
10070175 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
describe Multivariate::Helper do
before(:each) do
REDIS.flushall
Multivariate::Helper.current_user = {}
end
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Improved redis configuration by copying from resque
<DFF> @@ -2,7 +2,7 @@ require 'spec_helper'
describe Multivariate::Helper do
before(:each) do
- REDIS.flushall
+ Multivariate.redis.flushall
Multivariate::Helper.current_user = {}
end
| 1 | Improved redis configuration by copying from resque | 1 | .rb | rb | mit | splitrb/split |
10070176 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> README: clarify multiple goals
This confused us.
<DFF> @@ -559,12 +559,16 @@ Note that if you pass additional options, that should be a separate hash:
finished({ link_color: "purchase" }, reset: false)
```
-**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
+**NOTE:** This does not mean that a single experiment can complete more than one goal.
+
+Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
+**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
+
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
| 5 | README: clarify multiple goals | 1 | .md | md | mit | splitrb/split |
10070177 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> README: clarify multiple goals
This confused us.
<DFF> @@ -559,12 +559,16 @@ Note that if you pass additional options, that should be a separate hash:
finished({ link_color: "purchase" }, reset: false)
```
-**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
+**NOTE:** This does not mean that a single experiment can complete more than one goal.
+
+Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
+**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
+
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
| 5 | README: clarify multiple goals | 1 | .md | md | mit | splitrb/split |
10070178 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> README: clarify multiple goals
This confused us.
<DFF> @@ -559,12 +559,16 @@ Note that if you pass additional options, that should be a separate hash:
finished({ link_color: "purchase" }, reset: false)
```
-**NOTE:** This does not mean that a single experiment can have/complete progressive goals.
+**NOTE:** This does not mean that a single experiment can complete more than one goal.
+
+Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
+**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
+
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
| 5 | README: clarify multiple goals | 1 | .md | md | mit | splitrb/split |
10070179 | <NME> experiment.rb
<BEF> # frozen_string_literal: true
require 'rubystats'
module Split
class Experiment
attr_accessor :name
attr_accessor :goals
attr_accessor :alternative_probabilities
attr_accessor :metadata
attr_reader :alternatives
attr_reader :resettable
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def self.finished_key(key)
"#{key}:finished"
end
def set_alternatives_and_options(options)
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
end
def extract_alternatives_from_options(options)
alts = options[:alternatives] || []
if alts.length == 1
if alts[0].is_a? Hash
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> Merge pull request #674 from splitrb/encapsulate_ruby_stats
Encapsulate Split::Algorithms at our own module to avoid explicit calling rubystats everywhere
<DFF> @@ -1,7 +1,5 @@
# frozen_string_literal: true
-require 'rubystats'
-
module Split
class Experiment
attr_accessor :name
@@ -363,7 +361,7 @@ module Split
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
- simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
+ simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
| 1 | Merge pull request #674 from splitrb/encapsulate_ruby_stats | 3 | .rb | rb | mit | splitrb/split |
10070180 | <NME> experiment.rb
<BEF> # frozen_string_literal: true
require 'rubystats'
module Split
class Experiment
attr_accessor :name
attr_accessor :goals
attr_accessor :alternative_probabilities
attr_accessor :metadata
attr_reader :alternatives
attr_reader :resettable
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def self.finished_key(key)
"#{key}:finished"
end
def set_alternatives_and_options(options)
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
end
def extract_alternatives_from_options(options)
alts = options[:alternatives] || []
if alts.length == 1
if alts[0].is_a? Hash
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> Merge pull request #674 from splitrb/encapsulate_ruby_stats
Encapsulate Split::Algorithms at our own module to avoid explicit calling rubystats everywhere
<DFF> @@ -1,7 +1,5 @@
# frozen_string_literal: true
-require 'rubystats'
-
module Split
class Experiment
attr_accessor :name
@@ -363,7 +361,7 @@ module Split
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
- simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
+ simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
| 1 | Merge pull request #674 from splitrb/encapsulate_ruby_stats | 3 | .rb | rb | mit | splitrb/split |
10070181 | <NME> experiment.rb
<BEF> # frozen_string_literal: true
require 'rubystats'
module Split
class Experiment
attr_accessor :name
attr_accessor :goals
attr_accessor :alternative_probabilities
attr_accessor :metadata
attr_reader :alternatives
attr_reader :resettable
DEFAULT_OPTIONS = {
resettable: true
}
def self.find(name)
Split.cache(:experiments, name) do
return unless Split.redis.exists?(name)
Experiment.new(name).tap { |exp| exp.load_from_redis }
end
end
def initialize(name, options = {})
options = DEFAULT_OPTIONS.merge(options)
@name = name.to_s
extract_alternatives_from_options(options)
end
def self.finished_key(key)
"#{key}:finished"
end
def set_alternatives_and_options(options)
options_with_defaults = DEFAULT_OPTIONS.merge(
options.reject { |k, v| v.nil? }
)
self.alternatives = options_with_defaults[:alternatives]
self.goals = options_with_defaults[:goals]
self.resettable = options_with_defaults[:resettable]
self.algorithm = options_with_defaults[:algorithm]
self.metadata = options_with_defaults[:metadata]
end
def extract_alternatives_from_options(options)
alts = options[:alternatives] || []
if alts.length == 1
if alts[0].is_a? Hash
alts = alts[0].map { |k, v| { k => v } }
end
end
if alts.empty?
exp_config = Split.configuration.experiment_for(name)
if exp_config
alts = load_alternatives_from_configuration
options[:goals] = Split::GoalsCollection.new(@name).load_from_configuration
options[:metadata] = load_metadata_from_configuration
options[:resettable] = exp_config[:resettable]
options[:algorithm] = exp_config[:algorithm]
end
end
options[:alternatives] = alts
set_alternatives_and_options(options)
# calculate probability that each alternative is the winner
@alternative_probabilities = {}
alts
end
def save
validate!
if new_record?
start unless Split.configuration.start_manually
persist_experiment_configuration
elsif experiment_configuration_has_changed?
reset unless Split.configuration.reset_manually
persist_experiment_configuration
end
redis.hmset(experiment_config_key, :resettable, resettable.to_s,
:algorithm, algorithm.to_s)
self
end
def validate!
if @alternatives.empty? && Split.configuration.experiment_for(@name).nil?
raise ExperimentNotFound.new("Experiment #{@name} not found")
end
@alternatives.each { |a| a.validate! }
goals_collection.validate!
end
def new_record?
ExperimentCatalog.find(name).nil?
end
def ==(obj)
self.name == obj.name
end
def [](name)
alternatives.find { |a| a.name == name }
end
def algorithm
@algorithm ||= Split.configuration.algorithm
end
def algorithm=(algorithm)
@algorithm = algorithm.is_a?(String) ? algorithm.constantize : algorithm
end
def resettable=(resettable)
@resettable = resettable.is_a?(String) ? resettable == "true" : resettable
end
def alternatives=(alts)
@alternatives = alts.map do |alternative|
if alternative.kind_of?(Split::Alternative)
alternative
else
Split::Alternative.new(alternative, @name)
end
end
end
def winner
Split.cache(:experiment_winner, name) do
experiment_winner = redis.hget(:experiment_winner, name)
if experiment_winner
Split::Alternative.new(experiment_winner, name)
else
nil
end
end
end
def has_winner?
return @has_winner if defined? @has_winner
@has_winner = !winner.nil?
end
def winner=(winner_name)
redis.hset(:experiment_winner, name, winner_name.to_s)
@has_winner = true
Split.configuration.on_experiment_winner_choose.call(self)
end
def participant_count
alternatives.inject(0) { |sum, a| sum + a.participant_count }
end
def control
alternatives.first
end
def reset_winner
redis.hdel(:experiment_winner, name)
@has_winner = false
Split::Cache.clear_key(@name)
end
def start
redis.hset(:experiment_start_times, @name, Time.now.to_i)
end
def start_time
Split.cache(:experiment_start_times, @name) do
t = redis.hget(:experiment_start_times, @name)
if t
# Check if stored time is an integer
if t =~ /^[-+]?[0-9]+$/
Time.at(t.to_i)
else
Time.parse(t)
end
end
end
end
def next_alternative
winner || random_alternative
end
def random_alternative
if alternatives.length > 1
algorithm.choose_alternative(self)
else
alternatives.first
end
end
def version
@version ||= (redis.get("#{name}:version").to_i || 0)
end
def increment_version
@version = redis.incr("#{name}:version")
end
def key
if version.to_i > 0
"#{name}:#{version}"
else
name
end
end
def goals_key
"#{name}:goals"
end
def finished_key
self.class.finished_key(key)
end
def metadata_key
"#{name}:metadata"
end
def resettable?
resettable
end
def reset
Split.configuration.on_before_experiment_reset.call(self)
Split::Cache.clear_key(@name)
alternatives.each(&:reset)
reset_winner
Split.configuration.on_experiment_reset.call(self)
increment_version
end
def delete
Split.configuration.on_before_experiment_delete.call(self)
if Split.configuration.start_manually
redis.hdel(:experiment_start_times, @name)
end
reset_winner
redis.srem(:experiments, name)
remove_experiment_cohorting
remove_experiment_configuration
Split.configuration.on_experiment_delete.call(self)
increment_version
end
def delete_metadata
redis.del(metadata_key)
end
def load_from_redis
exp_config = redis.hgetall(experiment_config_key)
options = {
resettable: exp_config["resettable"],
algorithm: exp_config["algorithm"],
alternatives: load_alternatives_from_redis,
goals: Split::GoalsCollection.new(@name).load_from_redis,
metadata: load_metadata_from_redis
}
set_alternatives_and_options(options)
end
def calc_winning_alternatives
# Cache the winning alternatives so we recalculate them once per the specified interval.
intervals_since_epoch =
Time.now.utc.to_i / Split.configuration.winning_alternative_recalculation_interval
if self.calc_time != intervals_since_epoch
if goals.empty?
self.estimate_winning_alternative
else
goals.each do |goal|
self.estimate_winning_alternative(goal)
end
end
self.calc_time = intervals_since_epoch
self.save
end
end
def estimate_winning_alternative(goal = nil)
# initialize a hash of beta distributions based on the alternatives' conversion rates
beta_params = calc_beta_params(goal)
winning_alternatives = []
Split.configuration.beta_probability_simulations.times do
# calculate simulated conversion rates from the beta distributions
simulated_cr_hash = calc_simulated_conversion_rates(beta_params)
winning_alternative = find_simulated_winner(simulated_cr_hash)
# push the winning pair to the winning_alternatives array
winning_alternatives.push(winning_alternative)
end
winning_counts = count_simulated_wins(winning_alternatives)
@alternative_probabilities = calc_alternative_probabilities(winning_counts, Split.configuration.beta_probability_simulations)
write_to_alternatives(goal)
self.save
end
def write_to_alternatives(goal = nil)
alternatives.each do |alternative|
alternative.set_p_winner(@alternative_probabilities[alternative], goal)
end
end
def calc_alternative_probabilities(winning_counts, number_of_simulations)
alternative_probabilities = {}
winning_counts.each do |alternative, wins|
alternative_probabilities[alternative] = wins / number_of_simulations.to_f
end
alternative_probabilities
end
def count_simulated_wins(winning_alternatives)
# initialize a hash to keep track of winning alternative in simulations
winning_counts = {}
alternatives.each do |alternative|
winning_counts[alternative] = 0
end
# count number of times each alternative won, calculate probabilities, place in hash
winning_alternatives.each do |alternative|
winning_counts[alternative] += 1
end
winning_counts
end
def find_simulated_winner(simulated_cr_hash)
# figure out which alternative had the highest simulated conversion rate
winning_pair = ["", 0.0]
simulated_cr_hash.each do |alternative, rate|
if rate > winning_pair[1]
winning_pair = [alternative, rate]
end
end
winner = winning_pair[0]
winner
end
def calc_simulated_conversion_rates(beta_params)
simulated_cr_hash = {}
# create a hash which has the conversion rate pulled from each alternative's beta distribution
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
simulated_cr_hash[alternative] = simulated_conversion_rate
end
simulated_cr_hash
end
def calc_beta_params(goal = nil)
beta_params = {}
alternatives.each do |alternative|
conversions = goal.nil? ? alternative.completed_count : alternative.completed_count(goal)
alpha = 1 + conversions
beta = 1 + alternative.participant_count - conversions
params = [alpha, beta]
beta_params[alternative] = params
end
beta_params
end
def calc_time=(time)
redis.hset(experiment_config_key, :calc_time, time)
end
def calc_time
redis.hget(experiment_config_key, :calc_time).to_i
end
def jstring(goal = nil)
js_id = if goal.nil?
name
else
name + "-" + goal
end
js_id.gsub("/", "--")
end
def cohorting_disabled?
@cohorting_disabled ||= begin
value = redis.hget(experiment_config_key, :cohorting)
value.nil? ? false : value.downcase == "true"
end
end
def disable_cohorting
@cohorting_disabled = true
redis.hset(experiment_config_key, :cohorting, true.to_s)
end
def enable_cohorting
@cohorting_disabled = false
redis.hset(experiment_config_key, :cohorting, false.to_s)
end
protected
def experiment_config_key
"experiment_configurations/#{@name}"
end
def load_metadata_from_configuration
Split.configuration.experiment_for(@name)[:metadata]
end
def load_metadata_from_redis
meta = redis.get(metadata_key)
JSON.parse(meta) unless meta.nil?
end
def load_alternatives_from_configuration
alts = Split.configuration.experiment_for(@name)[:alternatives]
raise ArgumentError, "Experiment configuration is missing :alternatives array" unless alts
if alts.is_a?(Hash)
alts.keys
else
alts.flatten
end
end
def load_alternatives_from_redis
alternatives = redis.lrange(@name, 0, -1)
alternatives.map do |alt|
alt = begin
JSON.parse(alt)
rescue
alt
end
Split::Alternative.new(alt, @name)
end
end
private
def redis
Split.redis
end
def redis_interface
RedisInterface.new
end
def persist_experiment_configuration
redis_interface.add_to_set(:experiments, name)
redis_interface.persist_list(name, @alternatives.map { |alt| { alt.name => alt.weight }.to_json })
goals_collection.save
if @metadata
redis.set(metadata_key, @metadata.to_json)
else
delete_metadata
end
end
def remove_experiment_configuration
@alternatives.each(&:delete)
goals_collection.delete
delete_metadata
redis.del(@name)
end
def experiment_configuration_has_changed?
existing_experiment = Experiment.find(@name)
existing_experiment.alternatives.map(&:to_s) != @alternatives.map(&:to_s) ||
existing_experiment.goals != @goals ||
existing_experiment.metadata != @metadata
end
def goals_collection
Split::GoalsCollection.new(@name, @goals)
end
def remove_experiment_cohorting
@cohorting_disabled = false
redis.hdel(experiment_config_key, :cohorting)
end
end
end
<MSG> Merge pull request #674 from splitrb/encapsulate_ruby_stats
Encapsulate Split::Algorithms at our own module to avoid explicit calling rubystats everywhere
<DFF> @@ -1,7 +1,5 @@
# frozen_string_literal: true
-require 'rubystats'
-
module Split
class Experiment
attr_accessor :name
@@ -363,7 +361,7 @@ module Split
beta_params.each do |alternative, params|
alpha = params[0]
beta = params[1]
- simulated_conversion_rate = Rubystats::BetaDistribution.new(alpha, beta).rng
+ simulated_conversion_rate = Split::Algorithms.beta_distribution_rng(alpha, beta)
simulated_cr_hash[alternative] = simulated_conversion_rate
end
| 1 | Merge pull request #674 from splitrb/encapsulate_ruby_stats | 3 | .rb | rb | mit | splitrb/split |
10070182 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
context 'when redis is not available' do
before(:each) do
Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
end
context 'and db_failover config option is turned off' do
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Fix failing tests on Rubinius (issue #52)
<DFF> @@ -319,7 +319,7 @@ describe Split::Helper do
context 'when redis is not available' do
before(:each) do
- Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
+ Split.stub(:redis).and_raise(Errno::ECONNREFUSED.new)
end
context 'and db_failover config option is turned off' do
| 1 | Fix failing tests on Rubinius (issue #52) | 1 | .rb | rb | mit | splitrb/split |
10070183 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
context 'when redis is not available' do
before(:each) do
Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
end
context 'and db_failover config option is turned off' do
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Fix failing tests on Rubinius (issue #52)
<DFF> @@ -319,7 +319,7 @@ describe Split::Helper do
context 'when redis is not available' do
before(:each) do
- Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
+ Split.stub(:redis).and_raise(Errno::ECONNREFUSED.new)
end
context 'and db_failover config option is turned off' do
| 1 | Fix failing tests on Rubinius (issue #52) | 1 | .rb | rb | mit | splitrb/split |
10070184 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
context 'when redis is not available' do
before(:each) do
Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
end
context 'and db_failover config option is turned off' do
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Fix failing tests on Rubinius (issue #52)
<DFF> @@ -319,7 +319,7 @@ describe Split::Helper do
context 'when redis is not available' do
before(:each) do
- Split.stub(:redis).and_raise(Errno::ECONNREFUSED)
+ Split.stub(:redis).and_raise(Errno::ECONNREFUSED.new)
end
context 'and db_failover config option is turned off' do
| 1 | Fix failing tests on Rubinius (issue #52) | 1 | .rb | rb | mit | splitrb/split |
10070185 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 4
PATCH = 5
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> 144867444998371 dual cookie redis split persistence adapter (#420)
* Use fakeredis for testing
* require fakeredis manually
* Add fakeredis as a dev dependency
* v.1.5.0
* Add Dual Redis(logged-in)/cookie(logged-out) persistence adapter
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 4
- PATCH = 5
+ MINOR = 5
+ PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 2 | 144867444998371 dual cookie redis split persistence adapter (#420) | 2 | .rb | rb | mit | splitrb/split |
10070186 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 4
PATCH = 5
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> 144867444998371 dual cookie redis split persistence adapter (#420)
* Use fakeredis for testing
* require fakeredis manually
* Add fakeredis as a dev dependency
* v.1.5.0
* Add Dual Redis(logged-in)/cookie(logged-out) persistence adapter
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 4
- PATCH = 5
+ MINOR = 5
+ PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 2 | 144867444998371 dual cookie redis split persistence adapter (#420) | 2 | .rb | rb | mit | splitrb/split |
10070187 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 4
PATCH = 5
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> 144867444998371 dual cookie redis split persistence adapter (#420)
* Use fakeredis for testing
* require fakeredis manually
* Add fakeredis as a dev dependency
* v.1.5.0
* Add Dual Redis(logged-in)/cookie(logged-out) persistence adapter
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 4
- PATCH = 5
+ MINOR = 5
+ PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 2 | 144867444998371 dual cookie redis split persistence adapter (#420) | 2 | .rb | rb | mit | splitrb/split |
10070188 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
end
describe 'metadata' do
before do
Split.configuration.experiments = {
:my_experiment => {
:alternatives => ["one", "two"],
:resettable => false,
:metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
}
}
end
it 'should be passed to helper block' do
@params = { 'ab_test' => { 'my_experiment' => 'one' } }
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |alternative, meta|
meta
end).to eq('Meta1')
end
it 'should pass empty hash to helper block if library disabled' do
Split.configure do |config|
config.enabled = false
end
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |_, meta|
meta
end).to eq({})
end
end
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled
ab_test must return metadata on error or if split is disabled/excluded user
<DFF> @@ -277,33 +277,63 @@ describe Split::Helper do
end
describe 'metadata' do
- before do
- Split.configuration.experiments = {
- :my_experiment => {
- :alternatives => ["one", "two"],
- :resettable => false,
- :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ context 'is defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ }
}
- }
- end
+ end
+
+ it 'should be passed to helper block' do
+ @params = { 'ab_test' => { 'my_experiment' => 'two' } }
+ expect(ab_test('my_experiment')).to eq 'two'
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq('Meta2')
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
- it 'should be passed to helper block' do
- @params = { 'ab_test' => { 'my_experiment' => 'one' } }
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |alternative, meta|
- meta
- end).to eq('Meta1')
+ expect(ab_test('my_experiment')).to eq 'one'
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq('Meta1')
+ end
end
- it 'should pass empty hash to helper block if library disabled' do
- Split.configure do |config|
- config.enabled = false
+ context 'is not defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => nil
+ }
+ }
end
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |_, meta|
- meta
- end).to eq({})
+ it 'should be passed to helper block' do
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq({})
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
+
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq({})
+ end
end
end
| 51 | Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled | 21 | .rb | rb | mit | splitrb/split |
10070189 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
end
describe 'metadata' do
before do
Split.configuration.experiments = {
:my_experiment => {
:alternatives => ["one", "two"],
:resettable => false,
:metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
}
}
end
it 'should be passed to helper block' do
@params = { 'ab_test' => { 'my_experiment' => 'one' } }
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |alternative, meta|
meta
end).to eq('Meta1')
end
it 'should pass empty hash to helper block if library disabled' do
Split.configure do |config|
config.enabled = false
end
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |_, meta|
meta
end).to eq({})
end
end
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled
ab_test must return metadata on error or if split is disabled/excluded user
<DFF> @@ -277,33 +277,63 @@ describe Split::Helper do
end
describe 'metadata' do
- before do
- Split.configuration.experiments = {
- :my_experiment => {
- :alternatives => ["one", "two"],
- :resettable => false,
- :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ context 'is defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ }
}
- }
- end
+ end
+
+ it 'should be passed to helper block' do
+ @params = { 'ab_test' => { 'my_experiment' => 'two' } }
+ expect(ab_test('my_experiment')).to eq 'two'
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq('Meta2')
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
- it 'should be passed to helper block' do
- @params = { 'ab_test' => { 'my_experiment' => 'one' } }
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |alternative, meta|
- meta
- end).to eq('Meta1')
+ expect(ab_test('my_experiment')).to eq 'one'
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq('Meta1')
+ end
end
- it 'should pass empty hash to helper block if library disabled' do
- Split.configure do |config|
- config.enabled = false
+ context 'is not defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => nil
+ }
+ }
end
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |_, meta|
- meta
- end).to eq({})
+ it 'should be passed to helper block' do
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq({})
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
+
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq({})
+ end
end
end
| 51 | Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled | 21 | .rb | rb | mit | splitrb/split |
10070190 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
end
describe 'metadata' do
before do
Split.configuration.experiments = {
:my_experiment => {
:alternatives => ["one", "two"],
:resettable => false,
:metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
}
}
end
it 'should be passed to helper block' do
@params = { 'ab_test' => { 'my_experiment' => 'one' } }
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |alternative, meta|
meta
end).to eq('Meta1')
end
it 'should pass empty hash to helper block if library disabled' do
Split.configure do |config|
config.enabled = false
end
expect(ab_test('my_experiment')).to eq 'one'
expect(ab_test('my_experiment') do |_, meta|
meta
end).to eq({})
end
end
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled
ab_test must return metadata on error or if split is disabled/excluded user
<DFF> @@ -277,33 +277,63 @@ describe Split::Helper do
end
describe 'metadata' do
- before do
- Split.configuration.experiments = {
- :my_experiment => {
- :alternatives => ["one", "two"],
- :resettable => false,
- :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ context 'is defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => { 'one' => 'Meta1', 'two' => 'Meta2' }
+ }
}
- }
- end
+ end
+
+ it 'should be passed to helper block' do
+ @params = { 'ab_test' => { 'my_experiment' => 'two' } }
+ expect(ab_test('my_experiment')).to eq 'two'
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq('Meta2')
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
- it 'should be passed to helper block' do
- @params = { 'ab_test' => { 'my_experiment' => 'one' } }
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |alternative, meta|
- meta
- end).to eq('Meta1')
+ expect(ab_test('my_experiment')).to eq 'one'
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq('Meta1')
+ end
end
- it 'should pass empty hash to helper block if library disabled' do
- Split.configure do |config|
- config.enabled = false
+ context 'is not defined' do
+ before do
+ Split.configuration.experiments = {
+ :my_experiment => {
+ :alternatives => ["one", "two"],
+ :resettable => false,
+ :metadata => nil
+ }
+ }
end
- expect(ab_test('my_experiment')).to eq 'one'
- expect(ab_test('my_experiment') do |_, meta|
- meta
- end).to eq({})
+ it 'should be passed to helper block' do
+ expect(ab_test('my_experiment') do |alternative, meta|
+ meta
+ end).to eq({})
+ end
+
+ it 'should pass control metadata helper block if library disabled' do
+ Split.configure do |config|
+ config.enabled = false
+ end
+
+ expect(ab_test('my_experiment') do |_, meta|
+ meta
+ end).to eq({})
+ end
end
end
| 51 | Merge pull request #622 from splitrb/consistent-return-metadata-when-disabled | 21 | .rb | rb | mit | splitrb/split |
10070191 | <NME> dashboard.js
<BEF> function confirmReset() {
var agree=confirm("This will delete all data for this experiment?");
if (agree)
return true;
else
return false;
}
function confirmDelete() {
var agree=confirm("Are you sure you want to delete this experiment and all its data?");
if (agree)
return true;
else
return false;
}
function confirmWinner() {
var agree=confirm("This will now be returned for all users. Are you sure?");
if (agree)
return true;
else
return false;
}
function confirmReopen() {
var agree = confirm("This will reopen the experiment. Are you sure?");
return agree ? true : false;
}
function confirmEnableCohorting(){
var agree = confirm("This will enable the cohorting of the experiment. Are you sure?");
return agree ? true : false;
}
function confirmDisableCohorting(){
var agree = confirm("This will disable the cohorting of the experiment. Note: Existing participants will continue to receive their alternative and may continue to convert. Are you sure?");
return agree ? true : false;
}
<MSG> Merge pull request #188 from andrew/refactor-dashboard-js
Refactor dashboard js
<DFF> @@ -1,23 +1,19 @@
function confirmReset() {
- var agree=confirm("This will delete all data for this experiment?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will delete all data for this experiment?");
+ return agree ? true : false;
}
function confirmDelete() {
- var agree=confirm("Are you sure you want to delete this experiment and all its data?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("Are you sure you want to delete this experiment and all its data?");
+ return agree ? true : false;
}
function confirmWinner() {
- var agree=confirm("This will now be returned for all users. Are you sure?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will now be returned for all users. Are you sure?");
+ return agree ? true : false;
+}
+
+function confirmStep(step) {
+ var agree = confirm(step);
+ return agree ? true : false;
}
\ No newline at end of file
| 11 | Merge pull request #188 from andrew/refactor-dashboard-js | 15 | .js | js | mit | splitrb/split |
10070192 | <NME> dashboard.js
<BEF> function confirmReset() {
var agree=confirm("This will delete all data for this experiment?");
if (agree)
return true;
else
return false;
}
function confirmDelete() {
var agree=confirm("Are you sure you want to delete this experiment and all its data?");
if (agree)
return true;
else
return false;
}
function confirmWinner() {
var agree=confirm("This will now be returned for all users. Are you sure?");
if (agree)
return true;
else
return false;
}
function confirmReopen() {
var agree = confirm("This will reopen the experiment. Are you sure?");
return agree ? true : false;
}
function confirmEnableCohorting(){
var agree = confirm("This will enable the cohorting of the experiment. Are you sure?");
return agree ? true : false;
}
function confirmDisableCohorting(){
var agree = confirm("This will disable the cohorting of the experiment. Note: Existing participants will continue to receive their alternative and may continue to convert. Are you sure?");
return agree ? true : false;
}
<MSG> Merge pull request #188 from andrew/refactor-dashboard-js
Refactor dashboard js
<DFF> @@ -1,23 +1,19 @@
function confirmReset() {
- var agree=confirm("This will delete all data for this experiment?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will delete all data for this experiment?");
+ return agree ? true : false;
}
function confirmDelete() {
- var agree=confirm("Are you sure you want to delete this experiment and all its data?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("Are you sure you want to delete this experiment and all its data?");
+ return agree ? true : false;
}
function confirmWinner() {
- var agree=confirm("This will now be returned for all users. Are you sure?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will now be returned for all users. Are you sure?");
+ return agree ? true : false;
+}
+
+function confirmStep(step) {
+ var agree = confirm(step);
+ return agree ? true : false;
}
\ No newline at end of file
| 11 | Merge pull request #188 from andrew/refactor-dashboard-js | 15 | .js | js | mit | splitrb/split |
10070193 | <NME> dashboard.js
<BEF> function confirmReset() {
var agree=confirm("This will delete all data for this experiment?");
if (agree)
return true;
else
return false;
}
function confirmDelete() {
var agree=confirm("Are you sure you want to delete this experiment and all its data?");
if (agree)
return true;
else
return false;
}
function confirmWinner() {
var agree=confirm("This will now be returned for all users. Are you sure?");
if (agree)
return true;
else
return false;
}
function confirmReopen() {
var agree = confirm("This will reopen the experiment. Are you sure?");
return agree ? true : false;
}
function confirmEnableCohorting(){
var agree = confirm("This will enable the cohorting of the experiment. Are you sure?");
return agree ? true : false;
}
function confirmDisableCohorting(){
var agree = confirm("This will disable the cohorting of the experiment. Note: Existing participants will continue to receive their alternative and may continue to convert. Are you sure?");
return agree ? true : false;
}
<MSG> Merge pull request #188 from andrew/refactor-dashboard-js
Refactor dashboard js
<DFF> @@ -1,23 +1,19 @@
function confirmReset() {
- var agree=confirm("This will delete all data for this experiment?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will delete all data for this experiment?");
+ return agree ? true : false;
}
function confirmDelete() {
- var agree=confirm("Are you sure you want to delete this experiment and all its data?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("Are you sure you want to delete this experiment and all its data?");
+ return agree ? true : false;
}
function confirmWinner() {
- var agree=confirm("This will now be returned for all users. Are you sure?");
- if (agree)
- return true;
- else
- return false;
+ var agree = confirm("This will now be returned for all users. Are you sure?");
+ return agree ? true : false;
+}
+
+function confirmStep(step) {
+ var agree = confirm(step);
+ return agree ? true : false;
}
\ No newline at end of file
| 11 | Merge pull request #188 from andrew/refactor-dashboard-js | 15 | .js | js | mit | splitrb/split |
10070194 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
end
def begin_experiment(experiment, alternative_name = nil)
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Merge pull request #374 from andreibondarev/remove_legacy_code
Remove legacy methods
<DFF> @@ -87,6 +87,7 @@ module Split
end
def begin_experiment(experiment, alternative_name = nil)
+ warn 'DEPRECATION WARNING: begin_experiment is deprecated and will be removed from Split 1.5.0'
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
| 1 | Merge pull request #374 from andreibondarev/remove_legacy_code | 0 | .rb | rb | mit | splitrb/split |
10070195 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
end
def begin_experiment(experiment, alternative_name = nil)
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Merge pull request #374 from andreibondarev/remove_legacy_code
Remove legacy methods
<DFF> @@ -87,6 +87,7 @@ module Split
end
def begin_experiment(experiment, alternative_name = nil)
+ warn 'DEPRECATION WARNING: begin_experiment is deprecated and will be removed from Split 1.5.0'
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
| 1 | Merge pull request #374 from andreibondarev/remove_legacy_code | 0 | .rb | rb | mit | splitrb/split |
10070196 | <NME> helper.rb
<BEF> # frozen_string_literal: true
module Split
module Helper
OVERRIDE_PARAM_NAME = "ab_test"
module_function
def ab_test(metric_descriptor, control = nil, *alternatives)
begin
experiment = ExperimentCatalog.find_or_initialize(metric_descriptor, control, *alternatives)
alternative = if Split.configuration.enabled && !exclude_visitor?
experiment.save
raise(Split::InvalidExperimentsFormatError) unless (Split.configuration.experiments || {}).fetch(experiment.name.to_sym, {})[:combined_experiments].nil?
trial = Trial.new(user: ab_user, experiment: experiment,
override: override_alternative(experiment.name), exclude: exclude_visitor?,
disabled: split_generically_disabled?)
alt = trial.choose!(self)
alt ? alt.name : nil
else
control_variable(experiment.control)
end
rescue Errno::ECONNREFUSED, Redis::BaseError, SocketError => e
raise(e) unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
if Split.configuration.db_failover_allow_parameter_override
alternative = override_alternative(experiment.name) if override_present?(experiment.name)
alternative = control_variable(experiment.control) if split_generically_disabled?
end
ensure
alternative ||= control_variable(experiment.control)
end
if block_given?
metadata = experiment.metadata[alternative] if experiment.metadata
yield(alternative, metadata || {})
else
alternative
end
end
def reset!(experiment)
ab_user.delete(experiment.key)
end
def finish_experiment(experiment, options = { reset: true })
return false if active_experiments[experiment.name].nil?
return true if experiment.has_winner?
should_reset = experiment.resettable? && options[:reset]
if ab_user[experiment.finished_key] && !should_reset
true
else
alternative_name = ab_user[experiment.key]
trial = Trial.new(
user: ab_user,
experiment: experiment,
alternative: alternative_name,
goals: options[:goals],
)
trial.complete!(self)
if should_reset
reset!(experiment)
else
ab_user[experiment.finished_key] = true
end
end
end
def ab_finished(metric_descriptor, options = { reset: true })
return if exclude_visitor? || Split.configuration.disabled?
metric_descriptor, goals = normalize_metric(metric_descriptor)
experiments = Metric.possible_experiments(metric_descriptor)
if experiments.any?
experiments.each do |experiment|
next if override_present?(experiment.key)
finish_experiment(experiment, options.merge(goals: goals))
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
end
def begin_experiment(experiment, alternative_name = nil)
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
experiments.each do |experiment|
alternative_name = ab_user[experiment.key]
if alternative_name
alternative = experiment.alternatives.find { |alt| alt.name == alternative_name }
alternative.record_extra_info(key, value) if alternative
end
end
end
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def ab_active_experiments
ab_user.active_experiments
rescue => e
raise unless Split.configuration.db_failover
Split.configuration.db_failover_on_db_error.call(e)
end
def override_present?(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative(experiment_name)
override_alternative_by_params(experiment_name) || override_alternative_by_cookies(experiment_name)
end
def override_alternative_by_params(experiment_name)
defined?(params) && params[OVERRIDE_PARAM_NAME] && params[OVERRIDE_PARAM_NAME][experiment_name]
end
def override_alternative_by_cookies(experiment_name)
return unless defined?(request)
if request.cookies && request.cookies.key?("split_override")
experiments = JSON.parse(request.cookies["split_override"]) rescue {}
experiments[experiment_name]
end
end
def split_generically_disabled?
defined?(params) && params["SPLIT_DISABLE"]
end
def ab_user
@ab_user ||= User.new(self)
end
def exclude_visitor?
defined?(request) && (instance_exec(request, &Split.configuration.ignore_filter) || is_ignored_ip_address? || is_robot? || is_preview?)
end
def is_robot?
defined?(request) && request.user_agent =~ Split.configuration.robot_regex
end
def is_preview?
defined?(request) && defined?(request.headers) && request.headers["x-purpose"] == "preview"
end
def is_ignored_ip_address?
return false if Split.configuration.ignore_ip_addresses.empty?
Split.configuration.ignore_ip_addresses.each do |ip|
return true if defined?(request) && (request.ip == ip || (ip.class == Regexp && request.ip =~ ip))
end
false
end
def active_experiments
ab_user.active_experiments
end
def normalize_metric(metric_descriptor)
if Hash === metric_descriptor
experiment_name = metric_descriptor.keys.first
goals = Array(metric_descriptor.values.first)
else
experiment_name = metric_descriptor
goals = []
end
return experiment_name, goals
end
def control_variable(control)
Hash === control ? control.keys.first.to_s : control.to_s
end
end
end
<MSG> Merge pull request #374 from andreibondarev/remove_legacy_code
Remove legacy methods
<DFF> @@ -87,6 +87,7 @@ module Split
end
def begin_experiment(experiment, alternative_name = nil)
+ warn 'DEPRECATION WARNING: begin_experiment is deprecated and will be removed from Split 1.5.0'
alternative_name ||= experiment.control.name
ab_user[experiment.key] = alternative_name
alternative_name
| 1 | Merge pull request #374 from andreibondarev/remove_legacy_code | 0 | .rb | rb | mit | splitrb/split |
10070197 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"wiki_uri" => "https://github.com/splitrb/split/wiki",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = ">= 2.5.0"
s.required_rubygems_version = ">= 2.0.0"
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Revert "Remove Sinatra Dependency for Split Dashboard"
<DFF> @@ -31,6 +31,7 @@ Gem::Specification.new do |s|
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
+ s.add_dependency 'sinatra', '>= 1.2.6'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
| 1 | Revert "Remove Sinatra Dependency for Split Dashboard" | 0 | .gemspec | gemspec | mit | splitrb/split |
10070198 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"wiki_uri" => "https://github.com/splitrb/split/wiki",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = ">= 2.5.0"
s.required_rubygems_version = ">= 2.0.0"
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Revert "Remove Sinatra Dependency for Split Dashboard"
<DFF> @@ -31,6 +31,7 @@ Gem::Specification.new do |s|
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
+ s.add_dependency 'sinatra', '>= 1.2.6'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
| 1 | Revert "Remove Sinatra Dependency for Split Dashboard" | 0 | .gemspec | gemspec | mit | splitrb/split |
10070199 | <NME> split.gemspec
<BEF> # -*- encoding: utf-8 -*-
# frozen_string_literal: true
$:.push File.expand_path("../lib", __FILE__)
require "split/version"
Gem::Specification.new do |s|
s.name = "split"
s.version = Split::VERSION
s.platform = Gem::Platform::RUBY
s.authors = ["Andrew Nesbitt"]
s.licenses = ["MIT"]
s.email = ["[email protected]"]
s.homepage = "https://github.com/splitrb/split"
s.summary = "Rack based split testing framework"
s.metadata = {
"homepage_uri" => "https://github.com/splitrb/split",
"changelog_uri" => "https://github.com/splitrb/split/blob/main/CHANGELOG.md",
"source_code_uri" => "https://github.com/splitrb/split",
"bug_tracker_uri" => "https://github.com/splitrb/split/issues",
"wiki_uri" => "https://github.com/splitrb/split/wiki",
"mailing_list_uri" => "https://groups.google.com/d/forum/split-ruby"
}
s.required_ruby_version = ">= 2.5.0"
s.required_rubygems_version = ">= 2.0.0"
s.files = `git ls-files`.split("\n")
s.test_files = `git ls-files -- {test,spec,features}/*`.split("\n")
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
s.add_development_dependency "simplecov", "~> 0.15"
s.add_development_dependency "rack-test", "~> 2.0"
s.add_development_dependency "rake", "~> 13"
s.add_development_dependency "rspec", "~> 3.7"
s.add_development_dependency "pry", "~> 0.10"
s.add_development_dependency "rails", ">= 5.0"
end
<MSG> Revert "Remove Sinatra Dependency for Split Dashboard"
<DFF> @@ -31,6 +31,7 @@ Gem::Specification.new do |s|
s.require_paths = ["lib"]
s.add_dependency 'redis', '>= 4.2'
+ s.add_dependency 'sinatra', '>= 1.2.6'
s.add_dependency 'rubystats', '>= 0.3.0'
s.add_development_dependency 'bundler', '>= 1.17'
| 1 | Revert "Remove Sinatra Dependency for Split Dashboard" | 0 | .gemspec | gemspec | mit | splitrb/split |
10070200 | <NME> encapsulated_helper.rb
<BEF> ADDFILE
<MSG> Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished
<DFF> @@ -0,0 +1,52 @@
+# Split's helper exposes all kinds of methods we don't want to
+# mix into our model classes.
+#
+# This module exposes only two methods
+# - ab_test and
+# - ab_test_finished
+# that can safely be mixed into any class.
+#
+# Passes the instance of the class that it's mixed into to the
+# Split persistence adapter as context.
+#
+module Split
+ module EncapsulatedHelper
+
+ class ContextShim
+ include Split::Helper
+ def initialize(context)
+ @context = context
+ end
+ def ab_user
+ @ab_user ||= Split::Persistence.adapter.new(@context)
+ end
+ end
+
+ def ab_test(*arguments)
+ ret = split_context_shim.ab_test(*arguments)
+ # TODO there must be a better way to pass a block straight
+ # through to the original ab_test
+ if block_given?
+ if defined?(capture) # a block in a rails view
+ block = Proc.new { yield(ret) }
+ concat(capture(ret, &block))
+ false
+ else
+ yield(ret)
+ end
+ else
+ ret
+ end
+ end
+
+ def ab_test_finished(*arguments)
+ split_context_shim.finished *arguments
+ end
+
+ private
+
+ def split_context_shim
+ @split_context_shim ||= ContextShim.new self
+ end
+ end
+end
| 52 | Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished | 0 | .rb | rb | mit | splitrb/split |
10070201 | <NME> encapsulated_helper.rb
<BEF> ADDFILE
<MSG> Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished
<DFF> @@ -0,0 +1,52 @@
+# Split's helper exposes all kinds of methods we don't want to
+# mix into our model classes.
+#
+# This module exposes only two methods
+# - ab_test and
+# - ab_test_finished
+# that can safely be mixed into any class.
+#
+# Passes the instance of the class that it's mixed into to the
+# Split persistence adapter as context.
+#
+module Split
+ module EncapsulatedHelper
+
+ class ContextShim
+ include Split::Helper
+ def initialize(context)
+ @context = context
+ end
+ def ab_user
+ @ab_user ||= Split::Persistence.adapter.new(@context)
+ end
+ end
+
+ def ab_test(*arguments)
+ ret = split_context_shim.ab_test(*arguments)
+ # TODO there must be a better way to pass a block straight
+ # through to the original ab_test
+ if block_given?
+ if defined?(capture) # a block in a rails view
+ block = Proc.new { yield(ret) }
+ concat(capture(ret, &block))
+ false
+ else
+ yield(ret)
+ end
+ else
+ ret
+ end
+ end
+
+ def ab_test_finished(*arguments)
+ split_context_shim.finished *arguments
+ end
+
+ private
+
+ def split_context_shim
+ @split_context_shim ||= ContextShim.new self
+ end
+ end
+end
| 52 | Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished | 0 | .rb | rb | mit | splitrb/split |
10070202 | <NME> encapsulated_helper.rb
<BEF> ADDFILE
<MSG> Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished
<DFF> @@ -0,0 +1,52 @@
+# Split's helper exposes all kinds of methods we don't want to
+# mix into our model classes.
+#
+# This module exposes only two methods
+# - ab_test and
+# - ab_test_finished
+# that can safely be mixed into any class.
+#
+# Passes the instance of the class that it's mixed into to the
+# Split persistence adapter as context.
+#
+module Split
+ module EncapsulatedHelper
+
+ class ContextShim
+ include Split::Helper
+ def initialize(context)
+ @context = context
+ end
+ def ab_user
+ @ab_user ||= Split::Persistence.adapter.new(@context)
+ end
+ end
+
+ def ab_test(*arguments)
+ ret = split_context_shim.ab_test(*arguments)
+ # TODO there must be a better way to pass a block straight
+ # through to the original ab_test
+ if block_given?
+ if defined?(capture) # a block in a rails view
+ block = Proc.new { yield(ret) }
+ concat(capture(ret, &block))
+ false
+ else
+ yield(ret)
+ end
+ else
+ ret
+ end
+ end
+
+ def ab_test_finished(*arguments)
+ split_context_shim.finished *arguments
+ end
+
+ private
+
+ def split_context_shim
+ @split_context_shim ||= ContextShim.new self
+ end
+ end
+end
| 52 | Add Split::EncapsulatedHelper that you can mix into Rails controllers and models that will only expose ab_test and ab_test_finished | 0 | .rb | rb | mit | splitrb/split |
10070203 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Document testing with RSpec (#495)
<DFF> @@ -150,6 +150,36 @@ In the event you want to disable all tests without having to know the individual
It is not required to send `SPLIT_DISABLE=false` to activate Split.
+To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+
+```ruby
+# Recommended path for this file is 'spec/support/split_helper.rb', and you will need to ensure it
+# is `require`-d by rails_helper.rb or spec_helper.rb
+module SplitHelper
+
+ # Usage:
+ #
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
+ def use_ab_test(alternatives_by_experiment)
+ allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
+ alternative =
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ end
+ end
+end
+
+RSpec.configure do |config|
+ # Make the `use_ab_test` method available to all specs:
+ config.include SplitHelper
+end
+```
+
+
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
| 30 | Document testing with RSpec (#495) | 0 | .md | md | mit | splitrb/split |
10070204 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Document testing with RSpec (#495)
<DFF> @@ -150,6 +150,36 @@ In the event you want to disable all tests without having to know the individual
It is not required to send `SPLIT_DISABLE=false` to activate Split.
+To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+
+```ruby
+# Recommended path for this file is 'spec/support/split_helper.rb', and you will need to ensure it
+# is `require`-d by rails_helper.rb or spec_helper.rb
+module SplitHelper
+
+ # Usage:
+ #
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
+ def use_ab_test(alternatives_by_experiment)
+ allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
+ alternative =
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ end
+ end
+end
+
+RSpec.configure do |config|
+ # Make the `use_ab_test` method available to all specs:
+ config.include SplitHelper
+end
+```
+
+
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
| 30 | Document testing with RSpec (#495) | 0 | .md | md | mit | splitrb/split |
10070205 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Document testing with RSpec (#495)
<DFF> @@ -150,6 +150,36 @@ In the event you want to disable all tests without having to know the individual
It is not required to send `SPLIT_DISABLE=false` to activate Split.
+To aid testing with RSpec, write `split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
+
+```ruby
+# Recommended path for this file is 'spec/support/split_helper.rb', and you will need to ensure it
+# is `require`-d by rails_helper.rb or spec_helper.rb
+module SplitHelper
+
+ # Usage:
+ #
+ # Force a specific experiment alternative to always be returned:
+ # use_ab_test(signup_form: "single_page")
+ #
+ # Force alternatives for multiple experiments:
+ # use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
+ #
+ def use_ab_test(alternatives_by_experiment)
+ allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment|
+ alternative =
+ alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
+ end
+ end
+end
+
+RSpec.configure do |config|
+ # Make the `use_ab_test` method available to all specs:
+ config.include SplitHelper
+end
+```
+
+
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
| 30 | Document testing with RSpec (#495) | 0 | .md | md | mit | splitrb/split |
10070206 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 5
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> v1.6.0
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 5
+ MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | v1.6.0 | 1 | .rb | rb | mit | splitrb/split |
10070207 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 5
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> v1.6.0
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 5
+ MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | v1.6.0 | 1 | .rb | rb | mit | splitrb/split |
10070208 | <NME> version.rb
<BEF> # frozen_string_literal: true
module Split
MAJOR = 1
MINOR = 5
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
<MSG> v1.6.0
<DFF> @@ -1,7 +1,7 @@
# frozen_string_literal: true
module Split
MAJOR = 1
- MINOR = 5
+ MINOR = 6
PATCH = 0
VERSION = [MAJOR, MINOR, PATCH].join('.')
end
| 1 | v1.6.0 | 1 | .rb | rb | mit | splitrb/split |
10070209 | <NME> dashboard_helpers_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/dashboard/helpers"
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
confidence_level(0.000000000000006).should eql('No Change')
end
end
end
it "should consider a z-score of 1.65 <= z < 1.96 as 90% confident" do
expect(confidence_level(1.65)).to eq("90% confidence")
expect(confidence_level(1.80)).to eq("90% confidence")
end
it "should consider a z-score of 1.96 <= z < 2.58 as 95% confident" do
expect(confidence_level(1.96)).to eq("95% confidence")
expect(confidence_level(2.00)).to eq("95% confidence")
end
it "should consider a z-score of z >= 2.58 as 99% confident" do
expect(confidence_level(2.58)).to eq("99% confidence")
expect(confidence_level(3.00)).to eq("99% confidence")
end
describe "#round" do
it "can round number strings" do
expect(round("3.1415")).to eq BigDecimal("3.14")
end
it "can round number strings for precsion" do
expect(round("3.1415", 1)).to eq BigDecimal("3.1")
end
it "can handle invalid number strings" do
expect(round("N/A")).to be_zero
end
end
end
end
<MSG> Fixed dashboard range error
<DFF> @@ -6,7 +6,7 @@ include Split::DashboardHelpers
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
- confidence_level(0.000000000000006).should eql('No Change')
+ confidence_level(Complex "2.222670197524858e-18-0.03629895899899249i").should eql('No Change')
end
end
end
\ No newline at end of file
| 1 | Fixed dashboard range error | 1 | .rb | rb | mit | splitrb/split |
10070210 | <NME> dashboard_helpers_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/dashboard/helpers"
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
confidence_level(0.000000000000006).should eql('No Change')
end
end
end
it "should consider a z-score of 1.65 <= z < 1.96 as 90% confident" do
expect(confidence_level(1.65)).to eq("90% confidence")
expect(confidence_level(1.80)).to eq("90% confidence")
end
it "should consider a z-score of 1.96 <= z < 2.58 as 95% confident" do
expect(confidence_level(1.96)).to eq("95% confidence")
expect(confidence_level(2.00)).to eq("95% confidence")
end
it "should consider a z-score of z >= 2.58 as 99% confident" do
expect(confidence_level(2.58)).to eq("99% confidence")
expect(confidence_level(3.00)).to eq("99% confidence")
end
describe "#round" do
it "can round number strings" do
expect(round("3.1415")).to eq BigDecimal("3.14")
end
it "can round number strings for precsion" do
expect(round("3.1415", 1)).to eq BigDecimal("3.1")
end
it "can handle invalid number strings" do
expect(round("N/A")).to be_zero
end
end
end
end
<MSG> Fixed dashboard range error
<DFF> @@ -6,7 +6,7 @@ include Split::DashboardHelpers
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
- confidence_level(0.000000000000006).should eql('No Change')
+ confidence_level(Complex "2.222670197524858e-18-0.03629895899899249i").should eql('No Change')
end
end
end
\ No newline at end of file
| 1 | Fixed dashboard range error | 1 | .rb | rb | mit | splitrb/split |
10070211 | <NME> dashboard_helpers_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/dashboard/helpers"
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
confidence_level(0.000000000000006).should eql('No Change')
end
end
end
it "should consider a z-score of 1.65 <= z < 1.96 as 90% confident" do
expect(confidence_level(1.65)).to eq("90% confidence")
expect(confidence_level(1.80)).to eq("90% confidence")
end
it "should consider a z-score of 1.96 <= z < 2.58 as 95% confident" do
expect(confidence_level(1.96)).to eq("95% confidence")
expect(confidence_level(2.00)).to eq("95% confidence")
end
it "should consider a z-score of z >= 2.58 as 99% confident" do
expect(confidence_level(2.58)).to eq("99% confidence")
expect(confidence_level(3.00)).to eq("99% confidence")
end
describe "#round" do
it "can round number strings" do
expect(round("3.1415")).to eq BigDecimal("3.14")
end
it "can round number strings for precsion" do
expect(round("3.1415", 1)).to eq BigDecimal("3.1")
end
it "can handle invalid number strings" do
expect(round("N/A")).to be_zero
end
end
end
end
<MSG> Fixed dashboard range error
<DFF> @@ -6,7 +6,7 @@ include Split::DashboardHelpers
describe Split::DashboardHelpers do
describe 'confidence_level' do
it 'should handle very small numbers' do
- confidence_level(0.000000000000006).should eql('No Change')
+ confidence_level(Complex "2.222670197524858e-18-0.03629895899899249i").should eql('No Change')
end
end
end
\ No newline at end of file
| 1 | Fixed dashboard range error | 1 | .rb | rb | mit | splitrb/split |
10070212 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
it "should be resettable by default" do
expect(experiment.resettable).to be_truthy
end
it "should save to redis" do
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
context "nested hash" do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
expect(same_experiment_again.version).to eq(1)
end
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Removes metadata key when it updated to nil
<DFF> @@ -151,10 +151,23 @@ describe Split::Experiment do
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
+ let(:meta) { { a: 'b' }}
+
+ before do
+ experiment.save
+ end
+
+ it "should delete the key when metadata is removed" do
+ experiment.metadata = nil
+ experiment.save
+
+ expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
+ end
+
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
+
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -164,7 +177,6 @@ describe Split::Experiment do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -436,6 +448,21 @@ describe Split::Experiment do
expect(same_experiment_again.version).to eq(1)
end
+ context "when metadata is changed" do
+ it "should increase version" do
+ experiment.save
+ experiment.metadata = { 'foo' => 'bar' }
+
+ expect { experiment.save }.to change { experiment.version }.by(1)
+ end
+
+ it "does not increase version" do
+ experiment.metadata = nil
+ experiment.save
+ expect { experiment.save }.to change { experiment.version }.by(0)
+ end
+ end
+
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
| 29 | Removes metadata key when it updated to nil | 2 | .rb | rb | mit | splitrb/split |
10070213 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
it "should be resettable by default" do
expect(experiment.resettable).to be_truthy
end
it "should save to redis" do
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
context "nested hash" do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
expect(same_experiment_again.version).to eq(1)
end
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Removes metadata key when it updated to nil
<DFF> @@ -151,10 +151,23 @@ describe Split::Experiment do
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
+ let(:meta) { { a: 'b' }}
+
+ before do
+ experiment.save
+ end
+
+ it "should delete the key when metadata is removed" do
+ experiment.metadata = nil
+ experiment.save
+
+ expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
+ end
+
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
+
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -164,7 +177,6 @@ describe Split::Experiment do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -436,6 +448,21 @@ describe Split::Experiment do
expect(same_experiment_again.version).to eq(1)
end
+ context "when metadata is changed" do
+ it "should increase version" do
+ experiment.save
+ experiment.metadata = { 'foo' => 'bar' }
+
+ expect { experiment.save }.to change { experiment.version }.by(1)
+ end
+
+ it "does not increase version" do
+ experiment.metadata = nil
+ experiment.save
+ expect { experiment.save }.to change { experiment.version }.by(0)
+ end
+ end
+
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
| 29 | Removes metadata key when it updated to nil | 2 | .rb | rb | mit | splitrb/split |
10070214 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
it "should be resettable by default" do
expect(experiment.resettable).to be_truthy
end
it "should save to redis" do
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
context "nested hash" do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
expect(same_experiment_again.version).to eq(1)
end
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Removes metadata key when it updated to nil
<DFF> @@ -151,10 +151,23 @@ describe Split::Experiment do
describe '#metadata' do
let(:experiment) { Split::Experiment.new('basket_text', :alternatives => ['Basket', "Cart"], :algorithm => Split::Algorithms::Whiplash, :metadata => meta) }
+ let(:meta) { { a: 'b' }}
+
+ before do
+ experiment.save
+ end
+
+ it "should delete the key when metadata is removed" do
+ experiment.metadata = nil
+ experiment.save
+
+ expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
+ end
+
context 'simple hash' do
let(:meta) { { 'basket' => 'a', 'cart' => 'b' } }
+
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -164,7 +177,6 @@ describe Split::Experiment do
context 'nested hash' do
let(:meta) { { 'basket' => { 'one' => 'two' }, 'cart' => 'b' } }
it "should persist metadata in redis" do
- experiment.save
e = Split::ExperimentCatalog.find('basket_text')
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
@@ -436,6 +448,21 @@ describe Split::Experiment do
expect(same_experiment_again.version).to eq(1)
end
+ context "when metadata is changed" do
+ it "should increase version" do
+ experiment.save
+ experiment.metadata = { 'foo' => 'bar' }
+
+ expect { experiment.save }.to change { experiment.version }.by(1)
+ end
+
+ it "does not increase version" do
+ experiment.metadata = nil
+ experiment.save
+ expect { experiment.save }.to change { experiment.version }.by(0)
+ end
+ end
+
context 'when experiment configuration is changed' do
let(:reset_manually) { false }
| 29 | Removes metadata key when it updated to nil | 2 | .rb | rb | mit | splitrb/split |
10070215 | <NME> jquery.meow.js
<BEF> (function ($, window) {
'use strict';
// Meow queue
var default_meow_area,
meows = {
queue: {},
add: function (meow) {
this.queue[meow.timestamp] = meow;
},
get: function (timestamp) {
return this.queue[timestamp];
},
remove: function (timestamp) {
delete this.queue[timestamp];
},
size: function () {
var timestamp,
size = 0;
for (timestamp in this.queue) {
if (this.queue.hasOwnProperty(timestamp)) { size += 1; }
}
return size;
}
},
// Meow constructor
Meow = function (options) {
var that = this;
this.timestamp = new Date().getTime(); // used to identify this meow and timeout
this.hovered = false; // whether mouse is over or not
if (typeof default_meow_area === 'undefined'
&& typeof options.container === 'undefined') {
default_meow_area = $(window.document.createElement('div'))
.attr({'id': ((new Date()).getTime()), 'class': 'meows'});
$('body').prepend(default_meow_area);
}
if (meows.size() <= 0) {
if (typeof options.beforeCreateFirst === 'function') {
options.beforeCreateFirst.call(that);
}
}
if (typeof options.container === 'string') {
this.container = $(options.container);
} else {
this.container = default_meow_area;
}
if (typeof options.title === 'string') {
this.title = options.title;
}
if (typeof options.message === 'string') {
this.message = options.message;
} else if (options.message instanceof $) {
if (options.message.is('input,textarea,select')) {
this.message = options.message.val();
} else {
this.message = options.message.text();
}
if (typeof this.title === 'undefined' && typeof options.message.attr('title') === 'string') {
this.title = options.message.attr('title');
}
}
if (typeof options.icon === 'string') {
this.icon = options.icon;
}
if (options.sticky) {
this.duration = Infinity;
} else {
this.duration = options.duration || 5000;
}
// Call callback if it's defined (this = meow object)
if (typeof options.beforeCreate === 'function') {
options.beforeCreate.call(that);
}
// Add the meow to the meow area
this.container.append($(window.document.createElement('div'))
.attr('id', 'meow-' + this.timestamp.toString())
.addClass('meow')
.html($(window.document.createElement('div')).addClass('inner').html(this.message))
.hide()
.fadeIn(400));
this.manifest = $('#meow-' + this.timestamp.toString());
// Add title if it's defined
if (typeof this.title === 'string') {
this.manifest.find('.inner').prepend(
$(window.document.createElement('h1')).text(this.title)
);
}
// Add icon if it's defined
if (typeof that.icon === 'string') {
this.manifest.find('.inner').prepend(
$(window.document.createElement('div')).addClass('icon').html(
$(window.document.createElement('img')).attr('src', this.icon)
)
);
}
// Add close button if the meow isn't uncloseable
// TODO: this close button needs to be much prettier
if (options.closeable !== false) {
this.manifest.find('.inner').prepend(
$(window.document.createElement('a'))
.addClass('close')
.html('×')
.attr('href', '#close-meow-' + that.timestamp)
.click(function (e) {
e.preventDefault();
that.destroy();
})
);
}
this.manifest.bind('mouseenter mouseleave', function (event) {
if (event.type === 'mouseleave') {
that.hovered = false;
that.manifest.removeClass('hover');
// Destroy the mow on mouseleave if it's timed out
if (that.timestamp + that.duration <= new Date().getTime()) {
that.destroy();
}
} else {
that.hovered = true;
that.manifest.addClass('hover');
}
});
// Add a timeout if the duration isn't Infinity
if (this.duration !== Infinity) {
this.timeout = window.setTimeout(function () {
// Make sure this meow hasn't already been destroyed
if (typeof meows.get(that.timestamp) !== 'undefined') {
// Call callback if it's defined (this = meow DOM element)
if (typeof options.onTimeout === 'function') {
options.onTimeout.call(that.manifest);
}
// Don't destroy if user is hovering over meow
if (that.hovered !== true && typeof that === 'object') {
that.destroy();
}
}
}, that.duration);
}
this.destroy = function () {
if (that.destroyed !== true) {
// Call callback if it's defined (this = meow DOM element)
if (typeof options.beforeDestroy === 'function') {
options.beforeDestroy.call(that.manifest);
}
that.manifest.find('.inner').fadeTo(400, 0, function () {
that.manifest.slideUp(function () {
that.manifest.remove();
that.destroyed = true;
meows.remove(that.timestamp);
if (typeof options.afterDestroy === 'function') {
options.afterDestroy.call(null);
}
if (meows.size() <= 0) {
if (default_meow_area instanceof $) {
default_meow_area.remove();
default_meow_area = undefined;
}
if (typeof options.afterDestroyLast === 'function') {
options.afterDestroyLast.call(null);
}
}
});
});
}
};
};
$.fn.meow = function (args) {
var meow = new Meow(args);
meows.add(meow);
return meow;
};
$.meow = $.fn.meow;
};
$.fn.meow = function (args) {
meows.add(new Meow(args));
};
$.meow = $.fn.meow;
}(jQuery, window));
<MSG> Merge pull request #14 from mjangda/patch-1
Return Meow object upon creation
<DFF> @@ -191,7 +191,9 @@
};
$.fn.meow = function (args) {
- meows.add(new Meow(args));
+ var meow = new Meow(args);
+ meows.add(meow);
+ return meow;
};
$.meow = $.fn.meow;
}(jQuery, window));
| 3 | Merge pull request #14 from mjangda/patch-1 | 1 | .js | meow | mit | zacstewart/Meow |
10070216 | <NME> jquery.meow.js
<BEF> (function ($, window) {
'use strict';
// Meow queue
var default_meow_area,
meows = {
queue: {},
add: function (meow) {
this.queue[meow.timestamp] = meow;
},
get: function (timestamp) {
return this.queue[timestamp];
},
remove: function (timestamp) {
delete this.queue[timestamp];
},
size: function () {
var timestamp,
size = 0;
for (timestamp in this.queue) {
if (this.queue.hasOwnProperty(timestamp)) { size += 1; }
}
return size;
}
},
// Meow constructor
Meow = function (options) {
var that = this;
this.timestamp = new Date().getTime(); // used to identify this meow and timeout
this.hovered = false; // whether mouse is over or not
if (typeof default_meow_area === 'undefined'
&& typeof options.container === 'undefined') {
default_meow_area = $(window.document.createElement('div'))
.attr({'id': ((new Date()).getTime()), 'class': 'meows'});
$('body').prepend(default_meow_area);
}
if (meows.size() <= 0) {
if (typeof options.beforeCreateFirst === 'function') {
options.beforeCreateFirst.call(that);
}
}
if (typeof options.container === 'string') {
this.container = $(options.container);
} else {
this.container = default_meow_area;
}
if (typeof options.title === 'string') {
this.title = options.title;
}
if (typeof options.message === 'string') {
this.message = options.message;
} else if (options.message instanceof $) {
if (options.message.is('input,textarea,select')) {
this.message = options.message.val();
} else {
this.message = options.message.text();
}
if (typeof this.title === 'undefined' && typeof options.message.attr('title') === 'string') {
this.title = options.message.attr('title');
}
}
if (typeof options.icon === 'string') {
this.icon = options.icon;
}
if (options.sticky) {
this.duration = Infinity;
} else {
this.duration = options.duration || 5000;
}
// Call callback if it's defined (this = meow object)
if (typeof options.beforeCreate === 'function') {
options.beforeCreate.call(that);
}
// Add the meow to the meow area
this.container.append($(window.document.createElement('div'))
.attr('id', 'meow-' + this.timestamp.toString())
.addClass('meow')
.html($(window.document.createElement('div')).addClass('inner').html(this.message))
.hide()
.fadeIn(400));
this.manifest = $('#meow-' + this.timestamp.toString());
// Add title if it's defined
if (typeof this.title === 'string') {
this.manifest.find('.inner').prepend(
$(window.document.createElement('h1')).text(this.title)
);
}
// Add icon if it's defined
if (typeof that.icon === 'string') {
this.manifest.find('.inner').prepend(
$(window.document.createElement('div')).addClass('icon').html(
$(window.document.createElement('img')).attr('src', this.icon)
)
);
}
// Add close button if the meow isn't uncloseable
// TODO: this close button needs to be much prettier
if (options.closeable !== false) {
this.manifest.find('.inner').prepend(
$(window.document.createElement('a'))
.addClass('close')
.html('×')
.attr('href', '#close-meow-' + that.timestamp)
.click(function (e) {
e.preventDefault();
that.destroy();
})
);
}
this.manifest.bind('mouseenter mouseleave', function (event) {
if (event.type === 'mouseleave') {
that.hovered = false;
that.manifest.removeClass('hover');
// Destroy the mow on mouseleave if it's timed out
if (that.timestamp + that.duration <= new Date().getTime()) {
that.destroy();
}
} else {
that.hovered = true;
that.manifest.addClass('hover');
}
});
// Add a timeout if the duration isn't Infinity
if (this.duration !== Infinity) {
this.timeout = window.setTimeout(function () {
// Make sure this meow hasn't already been destroyed
if (typeof meows.get(that.timestamp) !== 'undefined') {
// Call callback if it's defined (this = meow DOM element)
if (typeof options.onTimeout === 'function') {
options.onTimeout.call(that.manifest);
}
// Don't destroy if user is hovering over meow
if (that.hovered !== true && typeof that === 'object') {
that.destroy();
}
}
}, that.duration);
}
this.destroy = function () {
if (that.destroyed !== true) {
// Call callback if it's defined (this = meow DOM element)
if (typeof options.beforeDestroy === 'function') {
options.beforeDestroy.call(that.manifest);
}
that.manifest.find('.inner').fadeTo(400, 0, function () {
that.manifest.slideUp(function () {
that.manifest.remove();
that.destroyed = true;
meows.remove(that.timestamp);
if (typeof options.afterDestroy === 'function') {
options.afterDestroy.call(null);
}
if (meows.size() <= 0) {
if (default_meow_area instanceof $) {
default_meow_area.remove();
default_meow_area = undefined;
}
if (typeof options.afterDestroyLast === 'function') {
options.afterDestroyLast.call(null);
}
}
});
});
}
};
};
$.fn.meow = function (args) {
var meow = new Meow(args);
meows.add(meow);
return meow;
};
$.meow = $.fn.meow;
};
$.fn.meow = function (args) {
meows.add(new Meow(args));
};
$.meow = $.fn.meow;
}(jQuery, window));
<MSG> Merge pull request #14 from mjangda/patch-1
Return Meow object upon creation
<DFF> @@ -191,7 +191,9 @@
};
$.fn.meow = function (args) {
- meows.add(new Meow(args));
+ var meow = new Meow(args);
+ meows.add(meow);
+ return meow;
};
$.meow = $.fn.meow;
}(jQuery, window));
| 3 | Merge pull request #14 from mjangda/patch-1 | 1 | .js | meow | mit | zacstewart/Meow |
10070217 | <NME> alternative_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/alternative"
describe Split::Alternative do
let(:alternative) {
Split::Alternative.new("Basket", "basket_text")
}
let(:alternative2) {
Split::Alternative.new("Cart", "basket_text")
}
let!(:experiment) {
Split::ExperimentCatalog.find_or_create({ "basket_text" => ["purchase", "refund"] }, "Basket", "Cart")
}
let(:goal1) { "purchase" }
let(:goal2) { "refund" }
it "should have goals" do
expect(alternative.goals).to eq(["purchase", "refund"])
end
it "should have and only return the name" do
expect(alternative.name).to eq("Basket")
end
describe "weights" do
it "should set the weights" do
experiment = Split::Experiment.new("basket_text", alternatives: [{ "Basket" => 0.6 }, { "Cart" => 0.4 }])
first = experiment.alternatives[0]
expect(first.name).to eq("Basket")
expect(first.weight).to eq(0.6)
second = experiment.alternatives[1]
expect(second.name).to eq("Cart")
expect(second.weight).to eq(0.4)
end
it "accepts probability on alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
]
}
}
experiment = Split::Experiment.new(:my_experiment)
first = experiment.alternatives[0]
expect(first.name).to eq("control_opt")
expect(first.weight).to eq(0.67)
second = experiment.alternatives[1]
expect(second.name).to eq("second_opt")
expect(second.weight).to eq(0.1)
end
it "accepts probability on some alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.34],
["second_opt", 0.215],
["third_opt", 0.23],
["fourth_opt", 0.215]
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
#
it "allows name param without probability" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.18],
["second_opt", 0.18],
["third_opt", 0.64],
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
end
it "should have a default participation count of 0" do
expect(alternative.participant_count).to eq(0)
end
it "should have a default completed count of 0 for each goal" do
expect(alternative.completed_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
end
it "should belong to an experiment" do
expect(alternative.experiment.name).to eq(experiment.name)
end
it "should save to redis" do
alternative.save
expect(Split.redis.exists?("basket_text:Basket")).to be true
end
it "should increment participation count" do
old_participant_count = alternative.participant_count
alternative.increment_participation
expect(alternative.participant_count).to eq(old_participant_count+1)
end
it "should increment completed count for each goal" do
old_default_completed_count = alternative.completed_count
old_completed_count_for_goal1 = alternative.completed_count(goal1)
old_completed_count_for_goal2 = alternative.completed_count(goal2)
alternative.increment_completion
alternative.increment_completion(goal1)
alternative.increment_completion(goal2)
expect(alternative.completed_count).to eq(old_default_completed_count+1)
expect(alternative.completed_count(goal1)).to eq(old_completed_count_for_goal1+1)
expect(alternative.completed_count(goal2)).to eq(old_completed_count_for_goal2+1)
end
it "can be reset" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(5, goal2)
alternative.set_completed_count(6)
alternative.reset
expect(alternative.participant_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
expect(alternative.completed_count).to eq(0)
end
it "should know if it is the control of an experiment" do
expect(alternative.control?).to be_truthy
expect(alternative2.control?).to be_falsey
end
describe "unfinished_count" do
it "should be difference between participant and completed counts" do
alternative.increment_participation
expect(alternative.unfinished_count).to eq(alternative.participant_count)
end
it "should return the correct unfinished_count" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(3, goal2)
alternative.set_completed_count(2)
expect(alternative.unfinished_count).to eq(1)
end
end
describe "conversion rate" do
it "should be 0 if there are no conversions" do
expect(alternative.completed_count).to eq(0)
expect(alternative.conversion_rate).to eq(0)
end
it "calculate conversion rate" do
expect(alternative).to receive(:participant_count).exactly(6).times.and_return(10)
expect(alternative).to receive(:completed_count).and_return(4)
expect(alternative.conversion_rate).to eq(0.4)
expect(alternative).to receive(:completed_count).with(goal1).and_return(5)
expect(alternative.conversion_rate(goal1)).to eq(0.5)
expect(alternative).to receive(:completed_count).with(goal2).and_return(6)
expect(alternative.conversion_rate(goal2)).to eq(0.6)
end
end
describe "probability winner" do
before do
experiment.calc_winning_alternatives
end
it "should have a probability of being the winning alternative (p_winner)" do
expect(alternative.p_winner).not_to be_nil
end
it "should have a probability of being the winner for each goal" do
expect(alternative.p_winner(goal1)).not_to be_nil
end
it "should be possible to set the p_winner" do
alternative.set_p_winner(0.5)
expect(alternative.p_winner).to eq(0.5)
end
it "should be possible to set the p_winner for each goal" do
alternative.set_p_winner(0.5, goal1)
expect(alternative.p_winner(goal1)).to eq(0.5)
end
end
describe "z score" do
it "should return an error string when the control has 0 people" do
expect(alternative2.z_score).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal1)).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal2)).to eq("Needs 30+ participants.")
end
it "should return an error string when the data is skewed or incomplete as per the np > 5 test" do
control = experiment.control
control.participant_count = 100
control.set_completed_count(50)
alternative2.participant_count = 50
alternative2.set_completed_count(1)
expect(alternative2.z_score).to eq("Needs 5+ conversions.")
end
it "should return a float for a z_score given proper data" do
control = experiment.control
control.participant_count = 120
control.set_completed_count(20)
alternative2.participant_count = 100
alternative2.set_completed_count(25)
expect(alternative2.z_score).to be_kind_of(Float)
expect(alternative2.z_score).to_not eq(0)
end
it "should correctly calculate a z_score given proper data" do
control = experiment.control
control.participant_count = 126
control.set_completed_count(89)
alternative2.participant_count = 142
alternative2.set_completed_count(119)
expect(alternative2.z_score.round(2)).to eq(2.58)
end
it "should be N/A for the control" do
control = experiment.control
expect(control.z_score).to eq("N/A")
expect(control.z_score(goal1)).to eq("N/A")
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
end
describe "extra_info" do
saved_recorded_info = { "key_1" => 1, "key_2" => "2" }
Split.redis.hset "#{alternative.experiment_name}:#{alternative.name}", "recorded_info", saved_recorded_info.to_json
extra_info = alternative.extra_info
expect(extra_info).to eql(saved_recorded_info)
end
end
describe "record_extra_info" do
it "saves key" do
alternative.record_extra_info("signup", 1)
expect(alternative.extra_info["signup"]).to eql(1)
end
it "adds value to saved key's value second argument is number" do
alternative.record_extra_info("signup", 1)
alternative.record_extra_info("signup", 2)
expect(alternative.extra_info["signup"]).to eql(3)
end
it "sets saved's key value to the second argument if it's a string" do
alternative.record_extra_info("signup", "Value 1")
expect(alternative.extra_info["signup"]).to eql("Value 1")
alternative.record_extra_info("signup", "Value 2")
expect(alternative.extra_info["signup"]).to eql("Value 2")
end
end
end
<MSG> Avoid z_score numeric exception for conversion rates >1 (#503)
At normal operation, it's not possible for an experiment
to have a `participant_count` > `completed_count`.
However, by manually using `increment_participation`
(for a custom use-case or by accident) you can arrive
on such a situation.
If that happens, `Zscore.calculate` produces an error that
propagates all the way up to the web view.
Solution:
In the cornercase of a conversion rates taking a value >1,
it makes no sense to calculate a zscore value at all.
An escape clause was added
<DFF> @@ -273,6 +273,18 @@ describe Split::Alternative do
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
+
+ it "should not blow up for Conversion Rates > 1" do
+ control = experiment.control
+ control.participant_count = 3474
+ control.set_completed_count(4244)
+
+ alternative2.participant_count = 3434
+ alternative2.set_completed_count(4358)
+
+ expect { control.z_score }.not_to raise_error
+ expect { alternative2.z_score }.not_to raise_error
+ end
end
describe "extra_info" do
| 12 | Avoid z_score numeric exception for conversion rates >1 (#503) | 0 | .rb | rb | mit | splitrb/split |
10070218 | <NME> alternative_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/alternative"
describe Split::Alternative do
let(:alternative) {
Split::Alternative.new("Basket", "basket_text")
}
let(:alternative2) {
Split::Alternative.new("Cart", "basket_text")
}
let!(:experiment) {
Split::ExperimentCatalog.find_or_create({ "basket_text" => ["purchase", "refund"] }, "Basket", "Cart")
}
let(:goal1) { "purchase" }
let(:goal2) { "refund" }
it "should have goals" do
expect(alternative.goals).to eq(["purchase", "refund"])
end
it "should have and only return the name" do
expect(alternative.name).to eq("Basket")
end
describe "weights" do
it "should set the weights" do
experiment = Split::Experiment.new("basket_text", alternatives: [{ "Basket" => 0.6 }, { "Cart" => 0.4 }])
first = experiment.alternatives[0]
expect(first.name).to eq("Basket")
expect(first.weight).to eq(0.6)
second = experiment.alternatives[1]
expect(second.name).to eq("Cart")
expect(second.weight).to eq(0.4)
end
it "accepts probability on alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
]
}
}
experiment = Split::Experiment.new(:my_experiment)
first = experiment.alternatives[0]
expect(first.name).to eq("control_opt")
expect(first.weight).to eq(0.67)
second = experiment.alternatives[1]
expect(second.name).to eq("second_opt")
expect(second.weight).to eq(0.1)
end
it "accepts probability on some alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.34],
["second_opt", 0.215],
["third_opt", 0.23],
["fourth_opt", 0.215]
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
#
it "allows name param without probability" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.18],
["second_opt", 0.18],
["third_opt", 0.64],
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
end
it "should have a default participation count of 0" do
expect(alternative.participant_count).to eq(0)
end
it "should have a default completed count of 0 for each goal" do
expect(alternative.completed_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
end
it "should belong to an experiment" do
expect(alternative.experiment.name).to eq(experiment.name)
end
it "should save to redis" do
alternative.save
expect(Split.redis.exists?("basket_text:Basket")).to be true
end
it "should increment participation count" do
old_participant_count = alternative.participant_count
alternative.increment_participation
expect(alternative.participant_count).to eq(old_participant_count+1)
end
it "should increment completed count for each goal" do
old_default_completed_count = alternative.completed_count
old_completed_count_for_goal1 = alternative.completed_count(goal1)
old_completed_count_for_goal2 = alternative.completed_count(goal2)
alternative.increment_completion
alternative.increment_completion(goal1)
alternative.increment_completion(goal2)
expect(alternative.completed_count).to eq(old_default_completed_count+1)
expect(alternative.completed_count(goal1)).to eq(old_completed_count_for_goal1+1)
expect(alternative.completed_count(goal2)).to eq(old_completed_count_for_goal2+1)
end
it "can be reset" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(5, goal2)
alternative.set_completed_count(6)
alternative.reset
expect(alternative.participant_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
expect(alternative.completed_count).to eq(0)
end
it "should know if it is the control of an experiment" do
expect(alternative.control?).to be_truthy
expect(alternative2.control?).to be_falsey
end
describe "unfinished_count" do
it "should be difference between participant and completed counts" do
alternative.increment_participation
expect(alternative.unfinished_count).to eq(alternative.participant_count)
end
it "should return the correct unfinished_count" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(3, goal2)
alternative.set_completed_count(2)
expect(alternative.unfinished_count).to eq(1)
end
end
describe "conversion rate" do
it "should be 0 if there are no conversions" do
expect(alternative.completed_count).to eq(0)
expect(alternative.conversion_rate).to eq(0)
end
it "calculate conversion rate" do
expect(alternative).to receive(:participant_count).exactly(6).times.and_return(10)
expect(alternative).to receive(:completed_count).and_return(4)
expect(alternative.conversion_rate).to eq(0.4)
expect(alternative).to receive(:completed_count).with(goal1).and_return(5)
expect(alternative.conversion_rate(goal1)).to eq(0.5)
expect(alternative).to receive(:completed_count).with(goal2).and_return(6)
expect(alternative.conversion_rate(goal2)).to eq(0.6)
end
end
describe "probability winner" do
before do
experiment.calc_winning_alternatives
end
it "should have a probability of being the winning alternative (p_winner)" do
expect(alternative.p_winner).not_to be_nil
end
it "should have a probability of being the winner for each goal" do
expect(alternative.p_winner(goal1)).not_to be_nil
end
it "should be possible to set the p_winner" do
alternative.set_p_winner(0.5)
expect(alternative.p_winner).to eq(0.5)
end
it "should be possible to set the p_winner for each goal" do
alternative.set_p_winner(0.5, goal1)
expect(alternative.p_winner(goal1)).to eq(0.5)
end
end
describe "z score" do
it "should return an error string when the control has 0 people" do
expect(alternative2.z_score).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal1)).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal2)).to eq("Needs 30+ participants.")
end
it "should return an error string when the data is skewed or incomplete as per the np > 5 test" do
control = experiment.control
control.participant_count = 100
control.set_completed_count(50)
alternative2.participant_count = 50
alternative2.set_completed_count(1)
expect(alternative2.z_score).to eq("Needs 5+ conversions.")
end
it "should return a float for a z_score given proper data" do
control = experiment.control
control.participant_count = 120
control.set_completed_count(20)
alternative2.participant_count = 100
alternative2.set_completed_count(25)
expect(alternative2.z_score).to be_kind_of(Float)
expect(alternative2.z_score).to_not eq(0)
end
it "should correctly calculate a z_score given proper data" do
control = experiment.control
control.participant_count = 126
control.set_completed_count(89)
alternative2.participant_count = 142
alternative2.set_completed_count(119)
expect(alternative2.z_score.round(2)).to eq(2.58)
end
it "should be N/A for the control" do
control = experiment.control
expect(control.z_score).to eq("N/A")
expect(control.z_score(goal1)).to eq("N/A")
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
end
describe "extra_info" do
saved_recorded_info = { "key_1" => 1, "key_2" => "2" }
Split.redis.hset "#{alternative.experiment_name}:#{alternative.name}", "recorded_info", saved_recorded_info.to_json
extra_info = alternative.extra_info
expect(extra_info).to eql(saved_recorded_info)
end
end
describe "record_extra_info" do
it "saves key" do
alternative.record_extra_info("signup", 1)
expect(alternative.extra_info["signup"]).to eql(1)
end
it "adds value to saved key's value second argument is number" do
alternative.record_extra_info("signup", 1)
alternative.record_extra_info("signup", 2)
expect(alternative.extra_info["signup"]).to eql(3)
end
it "sets saved's key value to the second argument if it's a string" do
alternative.record_extra_info("signup", "Value 1")
expect(alternative.extra_info["signup"]).to eql("Value 1")
alternative.record_extra_info("signup", "Value 2")
expect(alternative.extra_info["signup"]).to eql("Value 2")
end
end
end
<MSG> Avoid z_score numeric exception for conversion rates >1 (#503)
At normal operation, it's not possible for an experiment
to have a `participant_count` > `completed_count`.
However, by manually using `increment_participation`
(for a custom use-case or by accident) you can arrive
on such a situation.
If that happens, `Zscore.calculate` produces an error that
propagates all the way up to the web view.
Solution:
In the cornercase of a conversion rates taking a value >1,
it makes no sense to calculate a zscore value at all.
An escape clause was added
<DFF> @@ -273,6 +273,18 @@ describe Split::Alternative do
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
+
+ it "should not blow up for Conversion Rates > 1" do
+ control = experiment.control
+ control.participant_count = 3474
+ control.set_completed_count(4244)
+
+ alternative2.participant_count = 3434
+ alternative2.set_completed_count(4358)
+
+ expect { control.z_score }.not_to raise_error
+ expect { alternative2.z_score }.not_to raise_error
+ end
end
describe "extra_info" do
| 12 | Avoid z_score numeric exception for conversion rates >1 (#503) | 0 | .rb | rb | mit | splitrb/split |
10070219 | <NME> alternative_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "split/alternative"
describe Split::Alternative do
let(:alternative) {
Split::Alternative.new("Basket", "basket_text")
}
let(:alternative2) {
Split::Alternative.new("Cart", "basket_text")
}
let!(:experiment) {
Split::ExperimentCatalog.find_or_create({ "basket_text" => ["purchase", "refund"] }, "Basket", "Cart")
}
let(:goal1) { "purchase" }
let(:goal2) { "refund" }
it "should have goals" do
expect(alternative.goals).to eq(["purchase", "refund"])
end
it "should have and only return the name" do
expect(alternative.name).to eq("Basket")
end
describe "weights" do
it "should set the weights" do
experiment = Split::Experiment.new("basket_text", alternatives: [{ "Basket" => 0.6 }, { "Cart" => 0.4 }])
first = experiment.alternatives[0]
expect(first.name).to eq("Basket")
expect(first.weight).to eq(0.6)
second = experiment.alternatives[1]
expect(second.name).to eq("Cart")
expect(second.weight).to eq(0.4)
end
it "accepts probability on alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
]
}
}
experiment = Split::Experiment.new(:my_experiment)
first = experiment.alternatives[0]
expect(first.name).to eq("control_opt")
expect(first.weight).to eq(0.67)
second = experiment.alternatives[1]
expect(second.name).to eq("second_opt")
expect(second.weight).to eq(0.1)
end
it "accepts probability on some alternatives" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.34],
["second_opt", 0.215],
["third_opt", 0.23],
["fourth_opt", 0.215]
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
#
it "allows name param without probability" do
Split.configuration.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
}
experiment = Split::Experiment.new(:my_experiment)
alts = experiment.alternatives
[
["control_opt", 0.18],
["second_opt", 0.18],
["third_opt", 0.64],
].each do |h|
name, weight = h
alt = alts.shift
expect(alt.name).to eq(name)
expect(alt.weight).to eq(weight)
end
end
end
it "should have a default participation count of 0" do
expect(alternative.participant_count).to eq(0)
end
it "should have a default completed count of 0 for each goal" do
expect(alternative.completed_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
end
it "should belong to an experiment" do
expect(alternative.experiment.name).to eq(experiment.name)
end
it "should save to redis" do
alternative.save
expect(Split.redis.exists?("basket_text:Basket")).to be true
end
it "should increment participation count" do
old_participant_count = alternative.participant_count
alternative.increment_participation
expect(alternative.participant_count).to eq(old_participant_count+1)
end
it "should increment completed count for each goal" do
old_default_completed_count = alternative.completed_count
old_completed_count_for_goal1 = alternative.completed_count(goal1)
old_completed_count_for_goal2 = alternative.completed_count(goal2)
alternative.increment_completion
alternative.increment_completion(goal1)
alternative.increment_completion(goal2)
expect(alternative.completed_count).to eq(old_default_completed_count+1)
expect(alternative.completed_count(goal1)).to eq(old_completed_count_for_goal1+1)
expect(alternative.completed_count(goal2)).to eq(old_completed_count_for_goal2+1)
end
it "can be reset" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(5, goal2)
alternative.set_completed_count(6)
alternative.reset
expect(alternative.participant_count).to eq(0)
expect(alternative.completed_count(goal1)).to eq(0)
expect(alternative.completed_count(goal2)).to eq(0)
expect(alternative.completed_count).to eq(0)
end
it "should know if it is the control of an experiment" do
expect(alternative.control?).to be_truthy
expect(alternative2.control?).to be_falsey
end
describe "unfinished_count" do
it "should be difference between participant and completed counts" do
alternative.increment_participation
expect(alternative.unfinished_count).to eq(alternative.participant_count)
end
it "should return the correct unfinished_count" do
alternative.participant_count = 10
alternative.set_completed_count(4, goal1)
alternative.set_completed_count(3, goal2)
alternative.set_completed_count(2)
expect(alternative.unfinished_count).to eq(1)
end
end
describe "conversion rate" do
it "should be 0 if there are no conversions" do
expect(alternative.completed_count).to eq(0)
expect(alternative.conversion_rate).to eq(0)
end
it "calculate conversion rate" do
expect(alternative).to receive(:participant_count).exactly(6).times.and_return(10)
expect(alternative).to receive(:completed_count).and_return(4)
expect(alternative.conversion_rate).to eq(0.4)
expect(alternative).to receive(:completed_count).with(goal1).and_return(5)
expect(alternative.conversion_rate(goal1)).to eq(0.5)
expect(alternative).to receive(:completed_count).with(goal2).and_return(6)
expect(alternative.conversion_rate(goal2)).to eq(0.6)
end
end
describe "probability winner" do
before do
experiment.calc_winning_alternatives
end
it "should have a probability of being the winning alternative (p_winner)" do
expect(alternative.p_winner).not_to be_nil
end
it "should have a probability of being the winner for each goal" do
expect(alternative.p_winner(goal1)).not_to be_nil
end
it "should be possible to set the p_winner" do
alternative.set_p_winner(0.5)
expect(alternative.p_winner).to eq(0.5)
end
it "should be possible to set the p_winner for each goal" do
alternative.set_p_winner(0.5, goal1)
expect(alternative.p_winner(goal1)).to eq(0.5)
end
end
describe "z score" do
it "should return an error string when the control has 0 people" do
expect(alternative2.z_score).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal1)).to eq("Needs 30+ participants.")
expect(alternative2.z_score(goal2)).to eq("Needs 30+ participants.")
end
it "should return an error string when the data is skewed or incomplete as per the np > 5 test" do
control = experiment.control
control.participant_count = 100
control.set_completed_count(50)
alternative2.participant_count = 50
alternative2.set_completed_count(1)
expect(alternative2.z_score).to eq("Needs 5+ conversions.")
end
it "should return a float for a z_score given proper data" do
control = experiment.control
control.participant_count = 120
control.set_completed_count(20)
alternative2.participant_count = 100
alternative2.set_completed_count(25)
expect(alternative2.z_score).to be_kind_of(Float)
expect(alternative2.z_score).to_not eq(0)
end
it "should correctly calculate a z_score given proper data" do
control = experiment.control
control.participant_count = 126
control.set_completed_count(89)
alternative2.participant_count = 142
alternative2.set_completed_count(119)
expect(alternative2.z_score.round(2)).to eq(2.58)
end
it "should be N/A for the control" do
control = experiment.control
expect(control.z_score).to eq("N/A")
expect(control.z_score(goal1)).to eq("N/A")
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
end
describe "extra_info" do
saved_recorded_info = { "key_1" => 1, "key_2" => "2" }
Split.redis.hset "#{alternative.experiment_name}:#{alternative.name}", "recorded_info", saved_recorded_info.to_json
extra_info = alternative.extra_info
expect(extra_info).to eql(saved_recorded_info)
end
end
describe "record_extra_info" do
it "saves key" do
alternative.record_extra_info("signup", 1)
expect(alternative.extra_info["signup"]).to eql(1)
end
it "adds value to saved key's value second argument is number" do
alternative.record_extra_info("signup", 1)
alternative.record_extra_info("signup", 2)
expect(alternative.extra_info["signup"]).to eql(3)
end
it "sets saved's key value to the second argument if it's a string" do
alternative.record_extra_info("signup", "Value 1")
expect(alternative.extra_info["signup"]).to eql("Value 1")
alternative.record_extra_info("signup", "Value 2")
expect(alternative.extra_info["signup"]).to eql("Value 2")
end
end
end
<MSG> Avoid z_score numeric exception for conversion rates >1 (#503)
At normal operation, it's not possible for an experiment
to have a `participant_count` > `completed_count`.
However, by manually using `increment_participation`
(for a custom use-case or by accident) you can arrive
on such a situation.
If that happens, `Zscore.calculate` produces an error that
propagates all the way up to the web view.
Solution:
In the cornercase of a conversion rates taking a value >1,
it makes no sense to calculate a zscore value at all.
An escape clause was added
<DFF> @@ -273,6 +273,18 @@ describe Split::Alternative do
expect(control.z_score(goal1)).to eq('N/A')
expect(control.z_score(goal2)).to eq('N/A')
end
+
+ it "should not blow up for Conversion Rates > 1" do
+ control = experiment.control
+ control.participant_count = 3474
+ control.set_completed_count(4244)
+
+ alternative2.participant_count = 3434
+ alternative2.set_completed_count(4358)
+
+ expect { control.z_score }.not_to raise_error
+ expect { alternative2.z_score }.not_to raise_error
+ end
end
describe "extra_info" do
| 12 | Avoid z_score numeric exception for conversion rates >1 (#503) | 0 | .rb | rb | mit | splitrb/split |
10070220 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
context 'redis_url configuration [DEPRECATED]' do
it 'should warn on set and assign to #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis_url = 'example_url'
expect(@config.redis).to eq('example_url')
end
it 'should warn on get and return #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis = 'example_url'
expect(@config.redis_url).to eq('example_url')
end
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
expect(@config.persistence_cookie_domain).to eq(nil)
end
it "should allow the persistence cookie domain to be configured" do
@config.persistence_cookie_domain = ".acme.com"
expect(@config.persistence_cookie_domain).to eq(".acme.com")
end
end
end
<MSG> Merge pull request #631 from splitrb/remove-deprecated-redis-url
Remove redis_url impl. Deprecated on ~2.2
<DFF> @@ -212,20 +212,6 @@ describe Split::Configuration do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
- context 'redis_url configuration [DEPRECATED]' do
- it 'should warn on set and assign to #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis_url = 'example_url'
- expect(@config.redis).to eq('example_url')
- end
-
- it 'should warn on get and return #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis = 'example_url'
- expect(@config.redis_url).to eq('example_url')
- end
- end
-
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
| 0 | Merge pull request #631 from splitrb/remove-deprecated-redis-url | 14 | .rb | rb | mit | splitrb/split |
10070221 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
context 'redis_url configuration [DEPRECATED]' do
it 'should warn on set and assign to #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis_url = 'example_url'
expect(@config.redis).to eq('example_url')
end
it 'should warn on get and return #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis = 'example_url'
expect(@config.redis_url).to eq('example_url')
end
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
expect(@config.persistence_cookie_domain).to eq(nil)
end
it "should allow the persistence cookie domain to be configured" do
@config.persistence_cookie_domain = ".acme.com"
expect(@config.persistence_cookie_domain).to eq(".acme.com")
end
end
end
<MSG> Merge pull request #631 from splitrb/remove-deprecated-redis-url
Remove redis_url impl. Deprecated on ~2.2
<DFF> @@ -212,20 +212,6 @@ describe Split::Configuration do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
- context 'redis_url configuration [DEPRECATED]' do
- it 'should warn on set and assign to #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis_url = 'example_url'
- expect(@config.redis).to eq('example_url')
- end
-
- it 'should warn on get and return #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis = 'example_url'
- expect(@config.redis_url).to eq('example_url')
- end
- end
-
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
| 0 | Merge pull request #631 from splitrb/remove-deprecated-redis-url | 14 | .rb | rb | mit | splitrb/split |
10070222 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
context 'redis_url configuration [DEPRECATED]' do
it 'should warn on set and assign to #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis_url = 'example_url'
expect(@config.redis).to eq('example_url')
end
it 'should warn on get and return #redis' do
expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
@config.redis = 'example_url'
expect(@config.redis_url).to eq('example_url')
end
end
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
expect(@config.persistence_cookie_domain).to eq(nil)
end
it "should allow the persistence cookie domain to be configured" do
@config.persistence_cookie_domain = ".acme.com"
expect(@config.persistence_cookie_domain).to eq(".acme.com")
end
end
end
<MSG> Merge pull request #631 from splitrb/remove-deprecated-redis-url
Remove redis_url impl. Deprecated on ~2.2
<DFF> @@ -212,20 +212,6 @@ describe Split::Configuration do
expect(@config.normalized_experiments).to eq({:my_experiment=>{:alternatives=>[{"control_opt"=>0.67}, [{"second_opt"=>0.1}, {"third_opt"=>0.23}]]}})
end
- context 'redis_url configuration [DEPRECATED]' do
- it 'should warn on set and assign to #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis_url = 'example_url'
- expect(@config.redis).to eq('example_url')
- end
-
- it 'should warn on get and return #redis' do
- expect(@config).to receive(:warn).with(/\[DEPRECATED\]/) { nil }
- @config.redis = 'example_url'
- expect(@config.redis_url).to eq('example_url')
- end
- end
-
context "redis configuration" do
it "should default to local redis server" do
expect(@config.redis).to eq("redis://localhost:6379")
| 0 | Merge pull request #631 from splitrb/remove-deprecated-redis-url | 14 | .rb | rb | mit | splitrb/split |
10070223 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
old_redis_url = ENV["REDIS_URL"]
ENV.delete("REDIS_URL")
expect(Split::Configuration.new.redis).to eq("redis://localhost:6379")
ENV["REDIS_URL"] = old_redis_url
end
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
end
end
end
<MSG> Merge pull request #664 from joedelia/feature/set-cookie-domain
Adds config option for cookie domain
<DFF> @@ -246,4 +246,15 @@ describe Split::Configuration do
end
end
+ context "persistence cookie domain" do
+ it "should default to nil" do
+ expect(@config.persistence_cookie_domain).to eq(nil)
+ end
+
+ it "should allow the persistence cookie domain to be configured" do
+ @config.persistence_cookie_domain = '.acme.com'
+ expect(@config.persistence_cookie_domain).to eq('.acme.com')
+ end
+ end
+
end
| 11 | Merge pull request #664 from joedelia/feature/set-cookie-domain | 0 | .rb | rb | mit | splitrb/split |
10070224 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
old_redis_url = ENV["REDIS_URL"]
ENV.delete("REDIS_URL")
expect(Split::Configuration.new.redis).to eq("redis://localhost:6379")
ENV["REDIS_URL"] = old_redis_url
end
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
end
end
end
<MSG> Merge pull request #664 from joedelia/feature/set-cookie-domain
Adds config option for cookie domain
<DFF> @@ -246,4 +246,15 @@ describe Split::Configuration do
end
end
+ context "persistence cookie domain" do
+ it "should default to nil" do
+ expect(@config.persistence_cookie_domain).to eq(nil)
+ end
+
+ it "should allow the persistence cookie domain to be configured" do
+ @config.persistence_cookie_domain = '.acme.com'
+ expect(@config.persistence_cookie_domain).to eq('.acme.com')
+ end
+ end
+
end
| 11 | Merge pull request #664 from joedelia/feature/set-cookie-domain | 0 | .rb | rb | mit | splitrb/split |
10070225 | <NME> configuration_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
describe Split::Configuration do
before(:each) { @config = Split::Configuration.new }
it "should provide a default value for ignore_ip_addresses" do
expect(@config.ignore_ip_addresses).to eq([])
end
it "should provide default values for db failover" do
expect(@config.db_failover).to be_falsey
expect(@config.db_failover_on_db_error).to be_a Proc
end
it "should not allow multiple experiments by default" do
expect(@config.allow_multiple_experiments).to be_falsey
end
it "should be enabled by default" do
expect(@config.enabled).to be_truthy
end
it "disabled is the opposite of enabled" do
@config.enabled = false
expect(@config.disabled?).to be_truthy
end
it "should not store the overridden test group per default" do
expect(@config.store_override).to be_falsey
end
it "should provide a default pattern for robots" do
%w[Baidu Gigabot Googlebot libwww-perl lwp-trivial msnbot SiteUptime Slurp WordPress ZIBB ZyBorg YandexBot AdsBot-Google Wget curl bitlybot facebookexternalhit spider].each do |robot|
expect(@config.robot_regex).to match(robot)
end
expect(@config.robot_regex).to match("EventMachine HttpClient")
expect(@config.robot_regex).to match("libwww-perl/5.836")
expect(@config.robot_regex).to match("Pingdom.com_bot_version_1.4_(http://www.pingdom.com)")
expect(@config.robot_regex).to match(" - ")
end
it "should accept real UAs with the robot regexp" do
expect(@config.robot_regex).not_to match("Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.4) Gecko/20091017 SeaMonkey/2.0")
expect(@config.robot_regex).not_to match("Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; F-6.0SP2-20041109; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022; .NET CLR 1.1.4322; InfoPath.3)")
end
it "should allow adding a bot to the bot list" do
@config.bots["newbot"] = "An amazing test bot"
expect(@config.robot_regex).to match("newbot")
end
it "should use the session adapter for persistence by default" do
expect(@config.persistence).to eq(Split::Persistence::SessionAdapter)
end
it "should load a metric" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
it "should allow loading of experiment using experment_for" do
@config.experiments = { my_experiment: { alternatives: ["control_opt", "other_opt"], metric: :my_metric } }
expect(@config.experiment_for(:my_experiment)).to eq({ alternatives: ["control_opt", ["other_opt"]] })
end
context "when experiments are defined via YAML" do
context "as strings" do
context "in a basic configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- Control Opt
- Alt One
- Alt Two
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "in a configuration with metadata" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
metadata:
Control Opt:
text: 'Control Option'
Alt One:
text: 'Alternative One'
Alt Two:
text: 'Alternative Two'
resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should have metadata on the experiment" do
meta = @config.normalized_experiments[:my_experiment][:metadata]
expect(meta).to_not be nil
expect(meta["Control Opt"]["text"]).to eq("Control Option")
end
end
context "in a complex configuration" do
before do
experiments_yaml = <<-eos
my_experiment:
alternatives:
- name: Control Opt
percent: 67
- name: Alt One
percent: 10
- name: Alt Two
percent: 23
resettable: false
metric: my_metric
another_experiment:
alternatives:
- a
- b
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: [{ "Control Opt"=>0.67 },
[{ "Alt One"=>0.1 }, { "Alt Two"=>0.23 }]] }, another_experiment: { alternatives: ["a", ["b"]] } })
end
it "should recognize metrics" do
expect(@config.metrics).not_to be_nil
expect(@config.metrics.keys).to eq([:my_metric])
end
end
end
context "as symbols" do
context "with valid YAML" do
before do
experiments_yaml = <<-eos
:my_experiment:
:alternatives:
- Control Opt
- Alt One
- Alt Two
:resettable: false
eos
@config.experiments = YAML.load(experiments_yaml)
end
it "should normalize experiments" do
expect(@config.normalized_experiments).to eq({ my_experiment: { resettable: false, alternatives: ["Control Opt", ["Alt One", "Alt Two"]] } })
end
end
context "with invalid YAML" do
let(:yaml) { YAML.load(input) }
context "with an empty string" do
let(:input) { "" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
context "with just the YAML header" do
let(:input) { "---" }
it "should raise an error" do
expect { @config.experiments = yaml }.to raise_error(Split::InvalidExperimentsFormatError)
end
end
end
end
end
it "should normalize experiments" do
@config.experiments = {
my_experiment: {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
}
expect(@config.normalized_experiments).to eq({ my_experiment: { alternatives: [{ "control_opt"=>0.67 }, [{ "second_opt"=>0.1 }, { "third_opt"=>0.23 }]] } })
end
context "redis configuration" do
it "should default to local redis server" do
old_redis_url = ENV["REDIS_URL"]
ENV.delete("REDIS_URL")
expect(Split::Configuration.new.redis).to eq("redis://localhost:6379")
ENV["REDIS_URL"] = old_redis_url
end
it "should allow for redis url to be configured" do
@config.redis = "custom_redis_url"
expect(@config.redis).to eq("custom_redis_url")
end
context "provided REDIS_URL environment variable" do
it "should use the ENV variable" do
old_redis_url = ENV["REDIS_URL"]
ENV["REDIS_URL"] = "env_redis_url"
expect(Split::Configuration.new.redis).to eq("env_redis_url")
ENV["REDIS_URL"] = old_redis_url
end
end
end
context "persistence cookie length" do
it "should default to 1 year" do
expect(@config.persistence_cookie_length).to eq(31536000)
end
it "should allow the persistence cookie length to be configured" do
@config.persistence_cookie_length = 2592000
expect(@config.persistence_cookie_length).to eq(2592000)
end
end
context "persistence cookie domain" do
it "should default to nil" do
end
end
end
<MSG> Merge pull request #664 from joedelia/feature/set-cookie-domain
Adds config option for cookie domain
<DFF> @@ -246,4 +246,15 @@ describe Split::Configuration do
end
end
+ context "persistence cookie domain" do
+ it "should default to nil" do
+ expect(@config.persistence_cookie_domain).to eq(nil)
+ end
+
+ it "should allow the persistence cookie domain to be configured" do
+ @config.persistence_cookie_domain = '.acme.com'
+ expect(@config.persistence_cookie_domain).to eq('.acme.com')
+ end
+ end
+
end
| 11 | Merge pull request #664 from joedelia/feature/set-cookie-domain | 0 | .rb | rb | mit | splitrb/split |
10070226 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
Split.redis = split_config[Rails.env]
```
### Redis Caching
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Add note for caching only being available on v4.0+
<DFF> @@ -756,7 +756,7 @@ split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
-### Redis Caching
+### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
| 1 | Add note for caching only being available on v4.0+ | 1 | .md | md | mit | splitrb/split |
10070227 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
Split.redis = split_config[Rails.env]
```
### Redis Caching
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Add note for caching only being available on v4.0+
<DFF> @@ -756,7 +756,7 @@ split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
-### Redis Caching
+### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
| 1 | Add note for caching only being available on v4.0+ | 1 | .md | md | mit | splitrb/split |
10070228 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
```ruby
Split.configure do |config|
# bot config
config.robot_regex = /my_custom_robot_regex/ # or
config.bots['newbot'] = "Description for bot with 'newbot' user agent, which will be added to config.robot_regex for exclusion"
# IP config
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
Split.redis = split_config[Rails.env]
```
### Redis Caching
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Add note for caching only being available on v4.0+
<DFF> @@ -756,7 +756,7 @@ split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
-### Redis Caching
+### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load, but
| 1 | Add note for caching only being available on v4.0+ | 1 | .md | md | mit | splitrb/split |
10070229 | <NME> style.css
<BEF> ADDFILE
<MSG> Renamed Server to Dashboard
<DFF> @@ -0,0 +1,34 @@
+html { background:#efefef; font-family:Arial, Verdana, sans-serif; font-size:13px; }
+body { padding:0; margin:0; }
+
+.header { background:#000; padding:8px 5% 0 5%; border-bottom:1px solid #444;border-bottom:5px solid #0080FF;}
+.header h1 { color:#333; font-size:90%; font-weight:bold; margin-bottom:6px;}
+.header ul li { display:inline;}
+.header ul li a { color:#fff; text-decoration:none; margin-right:10px; display:inline-block; padding:8px; -webkit-border-top-right-radius:6px; -webkit-border-top-left-radius:6px; -moz-border-radius-topleft:6px; -moz-border-radius-topright:6px; }
+.header ul li a:hover { background:#333;}
+.header ul li.current a { background:#0080FF; font-weight:bold; color:#fff;}
+
+#main { padding:10px 5%; background:#fff; overflow:hidden; }
+#main .logo { float:right; margin:10px;}
+#main span.hl { background:#efefef; padding:2px;}
+#main h1 { margin:10px 0; font-size:190%; font-weight:bold; color:#0080FF;}
+#main h2 { margin:10px 0; font-size:130%;}
+#main table { width:100%; margin:10px 0;}
+#main table tr td, #main table tr th { border:1px solid #ccc; padding:6px;}
+#main table tr th { background:#efefef; color:#888; font-size:80%; font-weight:bold;}
+#main table tr td.no-data { text-align:center; padding:40px 0; color:#999; font-style:italic; font-size:130%;}
+#main a { color:#111;}
+#main p { margin:5px 0;}
+#main p.intro { margin-bottom:15px; font-size:85%; color:#999; margin-top:0; line-height:1.3;}
+#main h1.wi { margin-bottom:5px;}
+#main p.sub { font-size:95%; color:#999;}
+
+#main table.queues { width:40%;}
+
+#main table .totals td{ background:#eee; font-weight:bold; }
+
+#footer { padding:10px 5%; background:#efefef; color:#999; font-size:85%; line-height:1.5; border-top:5px solid #ccc; padding-top:10px;}
+#footer p a { color:#999;}
+
+
+
| 34 | Renamed Server to Dashboard | 0 | .css | css | mit | splitrb/split |
10070230 | <NME> style.css
<BEF> ADDFILE
<MSG> Renamed Server to Dashboard
<DFF> @@ -0,0 +1,34 @@
+html { background:#efefef; font-family:Arial, Verdana, sans-serif; font-size:13px; }
+body { padding:0; margin:0; }
+
+.header { background:#000; padding:8px 5% 0 5%; border-bottom:1px solid #444;border-bottom:5px solid #0080FF;}
+.header h1 { color:#333; font-size:90%; font-weight:bold; margin-bottom:6px;}
+.header ul li { display:inline;}
+.header ul li a { color:#fff; text-decoration:none; margin-right:10px; display:inline-block; padding:8px; -webkit-border-top-right-radius:6px; -webkit-border-top-left-radius:6px; -moz-border-radius-topleft:6px; -moz-border-radius-topright:6px; }
+.header ul li a:hover { background:#333;}
+.header ul li.current a { background:#0080FF; font-weight:bold; color:#fff;}
+
+#main { padding:10px 5%; background:#fff; overflow:hidden; }
+#main .logo { float:right; margin:10px;}
+#main span.hl { background:#efefef; padding:2px;}
+#main h1 { margin:10px 0; font-size:190%; font-weight:bold; color:#0080FF;}
+#main h2 { margin:10px 0; font-size:130%;}
+#main table { width:100%; margin:10px 0;}
+#main table tr td, #main table tr th { border:1px solid #ccc; padding:6px;}
+#main table tr th { background:#efefef; color:#888; font-size:80%; font-weight:bold;}
+#main table tr td.no-data { text-align:center; padding:40px 0; color:#999; font-style:italic; font-size:130%;}
+#main a { color:#111;}
+#main p { margin:5px 0;}
+#main p.intro { margin-bottom:15px; font-size:85%; color:#999; margin-top:0; line-height:1.3;}
+#main h1.wi { margin-bottom:5px;}
+#main p.sub { font-size:95%; color:#999;}
+
+#main table.queues { width:40%;}
+
+#main table .totals td{ background:#eee; font-weight:bold; }
+
+#footer { padding:10px 5%; background:#efefef; color:#999; font-size:85%; line-height:1.5; border-top:5px solid #ccc; padding-top:10px;}
+#footer p a { color:#999;}
+
+
+
| 34 | Renamed Server to Dashboard | 0 | .css | css | mit | splitrb/split |
10070231 | <NME> style.css
<BEF> ADDFILE
<MSG> Renamed Server to Dashboard
<DFF> @@ -0,0 +1,34 @@
+html { background:#efefef; font-family:Arial, Verdana, sans-serif; font-size:13px; }
+body { padding:0; margin:0; }
+
+.header { background:#000; padding:8px 5% 0 5%; border-bottom:1px solid #444;border-bottom:5px solid #0080FF;}
+.header h1 { color:#333; font-size:90%; font-weight:bold; margin-bottom:6px;}
+.header ul li { display:inline;}
+.header ul li a { color:#fff; text-decoration:none; margin-right:10px; display:inline-block; padding:8px; -webkit-border-top-right-radius:6px; -webkit-border-top-left-radius:6px; -moz-border-radius-topleft:6px; -moz-border-radius-topright:6px; }
+.header ul li a:hover { background:#333;}
+.header ul li.current a { background:#0080FF; font-weight:bold; color:#fff;}
+
+#main { padding:10px 5%; background:#fff; overflow:hidden; }
+#main .logo { float:right; margin:10px;}
+#main span.hl { background:#efefef; padding:2px;}
+#main h1 { margin:10px 0; font-size:190%; font-weight:bold; color:#0080FF;}
+#main h2 { margin:10px 0; font-size:130%;}
+#main table { width:100%; margin:10px 0;}
+#main table tr td, #main table tr th { border:1px solid #ccc; padding:6px;}
+#main table tr th { background:#efefef; color:#888; font-size:80%; font-weight:bold;}
+#main table tr td.no-data { text-align:center; padding:40px 0; color:#999; font-style:italic; font-size:130%;}
+#main a { color:#111;}
+#main p { margin:5px 0;}
+#main p.intro { margin-bottom:15px; font-size:85%; color:#999; margin-top:0; line-height:1.3;}
+#main h1.wi { margin-bottom:5px;}
+#main p.sub { font-size:95%; color:#999;}
+
+#main table.queues { width:40%;}
+
+#main table .totals td{ background:#eee; font-weight:bold; }
+
+#footer { padding:10px 5%; background:#efefef; color:#999; font-size:85%; line-height:1.5; border-top:5px solid #ccc; padding-top:10px;}
+#footer p a { color:#999;}
+
+
+
| 34 | Renamed Server to Dashboard | 0 | .css | css | mit | splitrb/split |
10070232 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric,
}
}
end
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Merge pull request #367 from hkliya/patch-2
Update README.md
<DFF> @@ -493,7 +493,7 @@ Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
- metric: :my_metric,
+ metric: :my_metric
}
}
end
| 1 | Merge pull request #367 from hkliya/patch-2 | 1 | .md | md | mit | splitrb/split |
10070233 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric,
}
}
end
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Merge pull request #367 from hkliya/patch-2
Update README.md
<DFF> @@ -493,7 +493,7 @@ Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
- metric: :my_metric,
+ metric: :my_metric
}
}
end
| 1 | Merge pull request #367 from hkliya/patch-2 | 1 | .md | md | mit | splitrb/split |
10070234 | <NME> README.md
<BEF> # [Split](https://libraries.io/rubygems/split)
[](http://badge.fury.io/rb/split)
[](https://codeclimate.com/github/splitrb/split)
[](https://codeclimate.com/github/splitrb/split/coverage)
[](https://github.com/RichardLitt/standard-readme)
[](https://www.codetriage.com/splitrb/split)
> 📈 The Rack Based A/B testing framework https://libraries.io/rubygems/split
Split is a rack based A/B testing framework designed to work with Rails, Sinatra or any other rack based app.
Split is heavily inspired by the [Abingo](https://github.com/ryanb/abingo) and [Vanity](https://github.com/assaf/vanity) Rails A/B testing plugins and [Resque](https://github.com/resque/resque) in its use of Redis.
Split is designed to be hacker friendly, allowing for maximum customisation and extensibility.
## Install
### Requirements
Split v4.0+ is currently tested with Ruby >= 2.5 and Rails >= 5.2.
If your project requires compatibility with Ruby 2.4.x or older Rails versions. You can try v3.0 or v0.8.0(for Ruby 1.9.3)
Split uses Redis as a datastore.
Split only supports Redis 4.0 or greater.
If you're on OS X, Homebrew is the simplest way to install Redis:
```bash
brew install redis
redis-server /usr/local/etc/redis.conf
```
You now have a Redis daemon running on port `6379`.
### Setup
```bash
gem install split
```
#### Rails
Adding `gem 'split'` to your Gemfile will autoload it when rails starts up, as long as you've configured Redis it will 'just work'.
#### Sinatra
To configure Sinatra with Split you need to enable sessions and mix in the helper methods. Add the following lines at the top of your Sinatra app:
```ruby
require 'split'
class MySinatraApp < Sinatra::Base
enable :sessions
helpers Split::Helper
get '/' do
...
end
```
## Usage
To begin your A/B test use the `ab_test` method, naming your experiment with the first argument and then the different alternatives which you wish to test on as the other arguments.
`ab_test` returns one of the alternatives, if a user has already seen that test they will get the same alternative as before, which you can use to split your code on.
It can be used to render different templates, show different text or any other case based logic.
`ab_finished` is used to make a completion of an experiment, or conversion.
Example: View
```erb
<% ab_test(:login_button, "/images/button1.jpg", "/images/button2.jpg") do |button_file| %>
<%= image_tag(button_file, alt: "Login!") %>
<% end %>
```
Example: Controller
```ruby
def register_new_user
# See what level of free points maximizes users' decision to buy replacement points.
@starter_points = ab_test(:new_user_free_points, '100', '200', '300')
end
```
Example: Conversion tracking (in a controller!)
```ruby
def buy_new_points
# some business logic
ab_finished(:new_user_free_points)
end
```
Example: Conversion tracking (in a view)
```erb
Thanks for signing up, dude! <% ab_finished(:signup_page_redesign) %>
```
You can find more examples, tutorials and guides on the [wiki](https://github.com/splitrb/split/wiki).
## Statistical Validity
Split has two options for you to use to determine which alternative is the best.
The first option (default on the dashboard) uses a z test (n>30) for the difference between your control and alternative conversion rates to calculate statistical significance. This test will tell you whether an alternative is better or worse than your control, but it will not distinguish between which alternative is the best in an experiment with multiple alternatives. Split will only tell you if your experiment is 90%, 95%, or 99% significant, and this test only works if you have more than 30 participants and 5 conversions for each branch.
As per this [blog post](https://www.evanmiller.org/how-not-to-run-an-ab-test.html) on the pitfalls of A/B testing, it is highly recommended that you determine your requisite sample size for each branch before running the experiment. Otherwise, you'll have an increased rate of false positives (experiments which show a significant effect where really there is none).
[Here](https://www.evanmiller.org/ab-testing/sample-size.html) is a sample size calculator for your convenience.
The second option uses simulations from a beta distribution to determine the probability that the given alternative is the winner compared to all other alternatives. You can view these probabilities by clicking on the drop-down menu labeled "Confidence." This option should be used when the experiment has more than just 1 control and 1 alternative. It can also be used for a simple, 2-alternative A/B test.
Calculating the beta-distribution simulations for a large number of experiments can be slow, so the results are cached. You can specify how often they should be recalculated (the default is once per day).
```ruby
Split.configure do |config|
config.winning_alternative_recalculation_interval = 3600 # 1 hour
end
```
## Extras
### Weighted alternatives
Perhaps you only want to show an alternative to 10% of your visitors because it is very experimental or not yet fully load tested.
To do this you can pass a weight with each alternative in the following ways:
```ruby
ab_test(:homepage_design, {'Old' => 18}, {'New' => 2})
ab_test(:homepage_design, 'Old', {'New' => 1.0/9})
ab_test(:homepage_design, {'Old' => 9}, 'New')
```
This will only show the new alternative to visitors 1 in 10 times, the default weight for an alternative is 1.
### Overriding alternatives
For development and testing, you may wish to force your app to always return an alternative.
You can do this by passing it as a parameter in the url.
If you have an experiment called `button_color` with alternatives called `red` and `blue` used on your homepage, a url such as:
http://myawesomesite.com?ab_test[button_color]=red
will always have red buttons. This won't be stored in your session or count towards to results, unless you set the `store_override` configuration option.
In the event you want to disable all tests without having to know the individual experiment names, add a `SPLIT_DISABLE` query parameter.
http://myawesomesite.com?SPLIT_DISABLE=true
It is not required to send `SPLIT_DISABLE=false` to activate Split.
### Rspec Helper
To aid testing with RSpec, write `spec/support/split_helper.rb` and call `use_ab_test(alternatives_by_experiment)` in your specs as instructed below:
```ruby
# Create a file with these contents at 'spec/support/split_helper.rb'
# and ensure it is `require`d in your rails_helper.rb or spec_helper.rb
module SplitHelper
# Force a specific experiment alternative to always be returned:
# use_ab_test(signup_form: "single_page")
#
# Force alternatives for multiple experiments:
# use_ab_test(signup_form: "single_page", pricing: "show_enterprise_prices")
#
def use_ab_test(alternatives_by_experiment)
allow_any_instance_of(Split::Helper).to receive(:ab_test) do |_receiver, experiment, &block|
variant = alternatives_by_experiment.fetch(experiment) { |key| raise "Unknown experiment '#{key}'" }
block.call(variant) unless block.nil?
variant
end
end
end
# Make the `use_ab_test` method available to all specs:
RSpec.configure do |config|
config.include SplitHelper
end
```
Now you can call `use_ab_test(alternatives_by_experiment)` in your specs, for example:
```ruby
it "registers using experimental signup" do
use_ab_test experiment_name: "alternative_name"
post "/signups"
...
end
```
### Starting experiments manually
By default new A/B tests will be active right after deployment. In case you would like to start new test a while after
the deploy, you can do it by setting the `start_manually` configuration option to `true`.
After choosing this option tests won't be started right after deploy, but after pressing the `Start` button in Split admin dashboard. If a test is deleted from the Split dashboard, then it can only be started after pressing the `Start` button whenever being re-initialized.
### Reset after completion
When a user completes a test their session is reset so that they may start the test again in the future.
To stop this behaviour you can pass the following option to the `ab_finished` method:
```ruby
ab_finished(:experiment_name, reset: false)
```
The user will then always see the alternative they started with.
Any old unfinished experiment key will be deleted from the user's data storage if the experiment had been removed or is over and a winner had been chosen. This allows a user to enroll into any new experiment in cases when the `allow_multiple_experiments` config option is set to `false`.
### Reset experiments manually
By default Split automatically resets the experiment whenever it detects the configuration for an experiment has changed (e.g. you call `ab_test` with different alternatives). You can prevent this by setting the option `reset_manually` to `true`.
You may want to do this when you want to change something, like the variants' names, the metadata about an experiment, etc. without resetting everything.
### Multiple experiments at once
By default Split will avoid users participating in multiple experiments at once. This means you are less likely to skew results by adding in more variation to your tests.
To stop this behaviour and allow users to participate in multiple experiments at once set the `allow_multiple_experiments` config option to true like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = true
end
```
This will allow the user to participate in any number of experiments and belong to any alternative in each experiment. This has the possible downside of a variation in one experiment influencing the outcome of another.
To address this, setting the `allow_multiple_experiments` config option to 'control' like so:
```ruby
Split.configure do |config|
config.allow_multiple_experiments = 'control'
end
```
For this to work, each and every experiment you define must have an alternative named 'control'. This will allow the user to participate in multiple experiments as long as the user belongs to the alternative 'control' in each experiment. As soon as the user belongs to an alternative named something other than 'control' the user may not participate in any more experiments. Calling ab_test(<other experiments>) will always return the first alternative without adding the user to that experiment.
### Experiment Persistence
Split comes with three built-in persistence adapters for storing users and the alternatives they've been given for each experiment.
By default Split will store the tests for each user in the session.
You can optionally configure Split to use a cookie, Redis, or any custom adapter of your choosing.
#### Cookies
```ruby
Split.configure do |config|
config.persistence = :cookie
end
```
When using the cookie persistence, Split stores data into an anonymous tracking cookie named 'split', which expires in 1 year. To change that, set the `persistence_cookie_length` in the configuration (unit of time in seconds).
```ruby
Split.configure do |config|
config.persistence = :cookie
config.persistence_cookie_length = 2592000 # 30 days
end
```
The data stored consists of the experiment name and the variants the user is in. Example: { "experiment_name" => "variant_a" }
__Note:__ Using cookies depends on `ActionDispatch::Cookies` or any identical API
#### Redis
Using Redis will allow ab_users to persist across sessions or machines.
```ruby
Split.configure do |config|
config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: -> (context) { context.current_user_id })
# Equivalent
# config.persistence = Split::Persistence::RedisAdapter.with_config(lookup_by: :current_user_id)
end
```
Options:
* `lookup_by`: method to invoke per request for uniquely identifying ab_users (mandatory configuration)
* `namespace`: separate namespace to store these persisted values (default "persistence")
* `expire_seconds`: sets TTL for user key. (if a user is in multiple experiments most recent update will reset TTL for all their assignments)
#### Dual Adapter
The Dual Adapter allows the use of different persistence adapters for logged-in and logged-out users. A common use case is to use Redis for logged-in users and Cookies for logged-out users.
```ruby
cookie_adapter = Split::Persistence::CookieAdapter
redis_adapter = Split::Persistence::RedisAdapter.with_config(
lookup_by: -> (context) { context.send(:current_user).try(:id) },
expire_seconds: 2592000)
Split.configure do |config|
config.persistence = Split::Persistence::DualAdapter.with_config(
logged_in: -> (context) { !context.send(:current_user).try(:id).nil? },
logged_in_adapter: redis_adapter,
logged_out_adapter: cookie_adapter)
config.persistence_cookie_length = 2592000 # 30 days
end
```
#### Custom Adapter
Your custom adapter needs to implement the same API as existing adapters.
See `Split::Persistence::CookieAdapter` or `Split::Persistence::SessionAdapter` for a starting point.
```ruby
Split.configure do |config|
config.persistence = YourCustomAdapterClass
end
```
### Trial Event Hooks
You can define methods that will be called at the same time as experiment
alternative participation and goal completion.
For example:
``` ruby
Split.configure do |config|
config.on_trial = :log_trial # run on every trial
config.on_trial_choose = :log_trial_choose # run on trials with new users only
config.on_trial_complete = :log_trial_complete
end
```
Set these attributes to a method name available in the same context as the
`ab_test` method. These methods should accept one argument, a `Trial` instance.
``` ruby
def log_trial(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_choose(trial)
logger.info "[new user] experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
def log_trial_complete(trial)
logger.info "experiment=%s alternative=%s user=%s complete=true" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
#### Views
If you are running `ab_test` from a view, you must define your event
hook callback as a
[helper_method](https://apidock.com/rails/AbstractController/Helpers/ClassMethods/helper_method)
in the controller:
``` ruby
helper_method :log_trial_choose
def log_trial_choose(trial)
logger.info "experiment=%s alternative=%s user=%s" %
[ trial.experiment.name, trial.alternative, current_user.id ]
end
```
### Experiment Hooks
You can assign a proc that will be called when an experiment is reset or deleted. You can use these hooks to call methods within your application to keep data related to experiments in sync with Split.
For example:
``` ruby
Split.configure do |config|
# after experiment reset or deleted
config.on_experiment_reset = -> (example) { # Do something on reset }
config.on_experiment_delete = -> (experiment) { # Do something else on delete }
# before experiment reset or deleted
config.on_before_experiment_reset = -> (example) { # Do something on reset }
config.on_before_experiment_delete = -> (experiment) { # Do something else on delete }
# after experiment winner had been set
config.on_experiment_winner_choose = -> (experiment) { # Do something on winner choose }
end
```
## Web Interface
Split comes with a Sinatra-based front end to get an overview of how your experiments are doing.
If you are running Rails 2: You can mount this inside your app using Rack::URLMap in your `config.ru`
```ruby
require 'split/dashboard'
run Rack::URLMap.new \
"/" => Your::App.new,
"/split" => Split::Dashboard.new
```
However, if you are using Rails 3 or higher: You can mount this inside your app routes by first adding this to the Gemfile:
```ruby
gem 'split', require: 'split/dashboard'
```
Then adding this to config/routes.rb
```ruby
mount Split::Dashboard, at: 'split'
```
You may want to password protect that page, you can do so with `Rack::Auth::Basic` (in your split initializer file)
```ruby
# Rails apps or apps that already depend on activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
ActiveSupport::SecurityUtils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
# Apps without activesupport
Split::Dashboard.use Rack::Auth::Basic do |username, password|
# Protect against timing attacks:
# - Use & (do not use &&) so that it doesn't short circuit.
# - Use digests to stop length information leaking
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(username), ::Digest::SHA256.hexdigest(ENV["SPLIT_USERNAME"])) &
Rack::Utils.secure_compare(::Digest::SHA256.hexdigest(password), ::Digest::SHA256.hexdigest(ENV["SPLIT_PASSWORD"]))
end
```
You can even use Devise or any other Warden-based authentication method to authorize users. Just replace `mount Split::Dashboard, :at => 'split'` in `config/routes.rb` with the following:
```ruby
match "/split" => Split::Dashboard, anchor: false, via: [:get, :post, :delete], constraints: -> (request) do
request.env['warden'].authenticated? # are we authenticated?
request.env['warden'].authenticate! # authenticate if not already
# or even check any other condition such as request.env['warden'].user.is_admin?
end
```
More information on this [here](https://steve.dynedge.co.uk/2011/12/09/controlling-access-to-routes-and-rack-apps-in-rails-3-with-devise-and-warden/)
### Screenshot
## Configuration
You can override the default configuration options of Split like so:
```ruby
Split.configure do |config|
config.db_failover = true # handle Redis errors gracefully
config.db_failover_on_db_error = -> (error) { Rails.logger.error(error.message) }
config.allow_multiple_experiments = true
config.enabled = true
config.persistence = Split::Persistence::SessionAdapter
#config.start_manually = false ## new test will have to be started manually from the admin panel. default false
#config.reset_manually = false ## if true, it never resets the experiment data, even if the configuration changes
config.include_rails_helper = true
config.redis = "redis://custom.redis.url:6380"
end
```
Split looks for the Redis host in the environment variable `REDIS_URL` then
defaults to `redis://localhost:6379` if not specified by configure block.
On platforms like Heroku, Split will use the value of `REDIS_PROVIDER` to
determine which env variable key to use when retrieving the host config. This
defaults to `REDIS_URL`.
### Filtering
In most scenarios you don't want to have AB-Testing enabled for web spiders, robots or special groups of users.
Split provides functionality to filter this based on a predefined, extensible list of bots, IP-lists or custom exclude logic.
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric,
}
}
end
config.ignore_ip_addresses << '81.19.48.130' # or regex: /81\.19\.48\.[0-9]+/
# or provide your own filter functionality, the default is proc{ |request| is_robot? || is_ignored_ip_address? || is_preview? }
config.ignore_filter = -> (request) { CustomExcludeLogic.excludes?(request) }
end
```
### Experiment configuration
Instead of providing the experiment options inline, you can store them
in a hash. This hash can control your experiment's alternatives, weights,
algorithm and if the experiment resets once finished:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
resettable: false
},
:my_second_experiment => {
algorithm: 'Split::Algorithms::Whiplash',
alternatives: [
{ name: "a", percent: 67 },
{ name: "b", percent: 33 }
]
}
}
end
```
You can also store your experiments in a YAML file:
```ruby
Split.configure do |config|
config.experiments = YAML.load_file "config/experiments.yml"
end
```
You can then define the YAML file like:
```yaml
my_first_experiment:
alternatives:
- a
- b
my_second_experiment:
alternatives:
- name: a
percent: 67
- name: b
percent: 33
resettable: false
```
This simplifies the calls from your code:
```ruby
ab_test(:my_first_experiment)
```
and:
```ruby
ab_finished(:my_first_experiment)
```
You can also add meta data for each experiment, which is very useful when you need more than an alternative name to change behaviour:
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metadata: {
"a" => {"text" => "Have a fantastic day"},
"b" => {"text" => "Don't get hit by a bus"}
}
}
}
end
```
```yaml
my_first_experiment:
alternatives:
- a
- b
metadata:
a:
text: "Have a fantastic day"
b:
text: "Don't get hit by a bus"
```
This allows for some advanced experiment configuration using methods like:
```ruby
trial.alternative.name # => "a"
trial.metadata['text'] # => "Have a fantastic day"
```
or in views:
```erb
<% ab_test("my_first_experiment") do |alternative, meta| %>
<%= alternative %>
<small><%= meta['text'] %></small>
<% end %>
```
The keys used in meta data should be Strings
#### Metrics
You might wish to track generic metrics, such as conversions, and use
those to complete multiple different experiments without adding more to
your code. You can use the configuration hash to do this, thanks to
the `:metric` option.
```ruby
Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
metric: :my_metric
}
}
end
```
Your code may then track a completion using the metric instead of
the experiment name:
```ruby
ab_finished(:my_metric)
```
You can also create a new metric by instantiating and saving a new Metric object.
```ruby
Split::Metric.new(:my_metric)
Split::Metric.save
```
#### Goals
You might wish to allow an experiment to have multiple, distinguishable goals.
The API to define goals for an experiment is this:
```ruby
ab_test({link_color: ["purchase", "refund"]}, "red", "blue")
```
or you can define them in a configuration file:
```ruby
Split.configure do |config|
config.experiments = {
link_color: {
alternatives: ["red", "blue"],
goals: ["purchase", "refund"]
}
}
end
```
To complete a goal conversion, you do it like:
```ruby
ab_finished(link_color: "purchase")
```
Note that if you pass additional options, that should be a separate hash:
```ruby
ab_finished({ link_color: "purchase" }, reset: false)
```
**NOTE:** This does not mean that a single experiment can complete more than one goal.
Once you finish one of the goals, the test is considered to be completed, and finishing the other goal will no longer register. (Assuming the test runs with `reset: false`.)
**Good Example**: Test if listing Plan A first result in more conversions to Plan A (goal: "plana_conversion") or Plan B (goal: "planb_conversion").
**Bad Example**: Test if button color increases conversion rate through multiple steps of a funnel. THIS WILL NOT WORK.
**Bad Example**: Test both how button color affects signup *and* how it affects login, at the same time. THIS WILL NOT WORK.
#### Combined Experiments
If you want to test how button color affects signup *and* how it affects login at the same time, use combined experiments.
Configure like so:
```ruby
Split.configuration.experiments = {
:button_color_experiment => {
:alternatives => ["blue", "green"],
:combined_experiments => ["button_color_on_signup", "button_color_on_login"]
}
}
```
Starting the combined test starts all combined experiments
```ruby
ab_combined_test(:button_color_experiment)
```
Finish each combined test as normal
```ruby
ab_finished(:button_color_on_login)
ab_finished(:button_color_on_signup)
```
**Additional Configuration**:
* Be sure to enable `allow_multiple_experiments`
* In Sinatra include the CombinedExperimentsHelper
```
helpers Split::CombinedExperimentsHelper
```
### DB failover solution
Due to the fact that Redis has no automatic failover mechanism, it's
possible to switch on the `db_failover` config option, so that `ab_test`
and `ab_finished` will not crash in case of a db failure. `ab_test` always
delivers alternative A (the first one) in that case.
It's also possible to set a `db_failover_on_db_error` callback (proc)
for example to log these errors via Rails.logger.
### Redis
You may want to change the Redis host and port Split connects to, or
set various other options at startup.
Split has a `redis` setter which can be given a string or a Redis
object. This means if you're already using Redis in your app, Split
can re-use the existing connection.
String: `Split.redis = 'redis://localhost:6379'`
Redis: `Split.redis = $redis`
For our rails app we have a `config/initializers/split.rb` file where
we load `config/split.yml` by hand and set the Redis information
appropriately.
Here's our `config/split.yml`:
```yml
development: redis://localhost:6379
test: redis://localhost:6379
staging: redis://redis1.example.com:6379
fi: redis://localhost:6379
production: redis://redis1.example.com:6379
```
And our initializer:
```ruby
split_config = YAML.load_file(Rails.root.join('config', 'split.yml'))
Split.redis = split_config[Rails.env]
```
### Redis Caching (v4.0+)
In some high-volume usage scenarios, Redis load can be incurred by repeated
fetches for fairly static data. Enabling caching will reduce this load.
```ruby
Split.configuration.cache = true
````
This currently caches:
- `Split::ExperimentCatalog.find`
- `Split::Experiment.start_time`
- `Split::Experiment.winner`
## Namespaces
If you're running multiple, separate instances of Split you may want
to namespace the keyspaces so they do not overlap. This is not unlike
the approach taken by many memcached clients.
This feature can be provided by the [redis-namespace](https://github.com/defunkt/redis-namespace)
library. To configure Split to use `Redis::Namespace`, do the following:
1. Add `redis-namespace` to your Gemfile:
```ruby
gem 'redis-namespace'
```
2. Configure `Split.redis` to use a `Redis::Namespace` instance (possible in an
initializer):
```ruby
redis = Redis.new(url: ENV['REDIS_URL']) # or whatever config you want
Split.redis = Redis::Namespace.new(:your_namespace, redis: redis)
```
## Outside of a Web Session
Split provides the Helper module to facilitate running experiments inside web sessions.
Alternatively, you can access the underlying Metric, Trial, Experiment and Alternative objects to
conduct experiments that are not tied to a web session.
```ruby
# create a new experiment
experiment = Split::ExperimentCatalog.find_or_create('color', 'red', 'blue')
# create a new trial
trial = Split::Trial.new(:experiment => experiment)
# run trial
trial.choose!
# get the result, returns either red or blue
trial.alternative.name
# if the goal has been achieved, increment the successful completions for this alternative.
if goal_achieved?
trial.complete!
end
```
## Algorithms
By default, Split ships with `Split::Algorithms::WeightedSample` that randomly selects from possible alternatives for a traditional a/b test.
It is possible to specify static weights to favor certain alternatives.
`Split::Algorithms::Whiplash` is an implementation of a [multi-armed bandit algorithm](http://stevehanov.ca/blog/index.php?id=132).
This algorithm will automatically weight the alternatives based on their relative performance,
choosing the better-performing ones more often as trials are completed.
`Split::Algorithms::BlockRandomization` is an algorithm that ensures equal
participation across all alternatives. This algorithm will choose the alternative
with the fewest participants. In the event of multiple minimum participant alternatives
(i.e. starting a new "Block") the algorithm will choose a random alternative from
those minimum participant alternatives.
Users may also write their own algorithms. The default algorithm may be specified globally in the configuration file, or on a per experiment basis using the experiments hash of the configuration file.
To change the algorithm globally for all experiments, use the following in your initializer:
```ruby
Split.configure do |config|
config.algorithm = Split::Algorithms::Whiplash
end
```
## Extensions
- [Split::Export](https://github.com/splitrb/split-export) - Easily export A/B test data out of Split.
- [Split::Analytics](https://github.com/splitrb/split-analytics) - Push test data to Google Analytics.
- [Split::Mongoid](https://github.com/MongoHQ/split-mongoid) - Store experiment data in mongoid (still uses redis).
- [Split::Cacheable](https://github.com/harrystech/split_cacheable) - Automatically create cache buckets per test.
- [Split::Counters](https://github.com/bernardkroes/split-counters) - Add counters per experiment and alternative.
- [Split::Cli](https://github.com/craigmcnamara/split-cli) - A CLI to trigger Split A/B tests.
## Screencast
Ryan bates has produced an excellent 10 minute screencast about split on the Railscasts site: [A/B Testing with Split](http://railscasts.com/episodes/331-a-b-testing-with-split)
## Blogposts
* [Recipe: A/B testing with KISSMetrics and the split gem](https://robots.thoughtbot.com/post/9595887299/recipe-a-b-testing-with-kissmetrics-and-the-split-gem)
* [Rails A/B testing with Split on Heroku](http://blog.nathanhumbert.com/2012/02/rails-ab-testing-with-split-on-heroku.html)
## Backers
Support us with a monthly donation and help us continue our activities. [[Become a backer](https://opencollective.com/split#backer)]
<a href="https://opencollective.com/split/backer/0/website" target="_blank"><img src="https://opencollective.com/split/backer/0/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/1/website" target="_blank"><img src="https://opencollective.com/split/backer/1/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/2/website" target="_blank"><img src="https://opencollective.com/split/backer/2/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/3/website" target="_blank"><img src="https://opencollective.com/split/backer/3/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/4/website" target="_blank"><img src="https://opencollective.com/split/backer/4/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/5/website" target="_blank"><img src="https://opencollective.com/split/backer/5/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/6/website" target="_blank"><img src="https://opencollective.com/split/backer/6/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/7/website" target="_blank"><img src="https://opencollective.com/split/backer/7/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/8/website" target="_blank"><img src="https://opencollective.com/split/backer/8/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/9/website" target="_blank"><img src="https://opencollective.com/split/backer/9/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/10/website" target="_blank"><img src="https://opencollective.com/split/backer/10/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/11/website" target="_blank"><img src="https://opencollective.com/split/backer/11/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/12/website" target="_blank"><img src="https://opencollective.com/split/backer/12/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/13/website" target="_blank"><img src="https://opencollective.com/split/backer/13/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/14/website" target="_blank"><img src="https://opencollective.com/split/backer/14/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/15/website" target="_blank"><img src="https://opencollective.com/split/backer/15/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/16/website" target="_blank"><img src="https://opencollective.com/split/backer/16/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/17/website" target="_blank"><img src="https://opencollective.com/split/backer/17/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/18/website" target="_blank"><img src="https://opencollective.com/split/backer/18/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/19/website" target="_blank"><img src="https://opencollective.com/split/backer/19/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/20/website" target="_blank"><img src="https://opencollective.com/split/backer/20/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/21/website" target="_blank"><img src="https://opencollective.com/split/backer/21/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/22/website" target="_blank"><img src="https://opencollective.com/split/backer/22/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/23/website" target="_blank"><img src="https://opencollective.com/split/backer/23/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/24/website" target="_blank"><img src="https://opencollective.com/split/backer/24/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/25/website" target="_blank"><img src="https://opencollective.com/split/backer/25/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/26/website" target="_blank"><img src="https://opencollective.com/split/backer/26/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/27/website" target="_blank"><img src="https://opencollective.com/split/backer/27/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/28/website" target="_blank"><img src="https://opencollective.com/split/backer/28/avatar.svg"></a>
<a href="https://opencollective.com/split/backer/29/website" target="_blank"><img src="https://opencollective.com/split/backer/29/avatar.svg"></a>
## Sponsors
Become a sponsor and get your logo on our README on Github with a link to your site. [[Become a sponsor](https://opencollective.com/split#sponsor)]
<a href="https://opencollective.com/split/sponsor/0/website" target="_blank"><img src="https://opencollective.com/split/sponsor/0/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/1/website" target="_blank"><img src="https://opencollective.com/split/sponsor/1/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/2/website" target="_blank"><img src="https://opencollective.com/split/sponsor/2/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/3/website" target="_blank"><img src="https://opencollective.com/split/sponsor/3/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/4/website" target="_blank"><img src="https://opencollective.com/split/sponsor/4/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/5/website" target="_blank"><img src="https://opencollective.com/split/sponsor/5/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/6/website" target="_blank"><img src="https://opencollective.com/split/sponsor/6/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/7/website" target="_blank"><img src="https://opencollective.com/split/sponsor/7/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/8/website" target="_blank"><img src="https://opencollective.com/split/sponsor/8/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/9/website" target="_blank"><img src="https://opencollective.com/split/sponsor/9/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/10/website" target="_blank"><img src="https://opencollective.com/split/sponsor/10/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/11/website" target="_blank"><img src="https://opencollective.com/split/sponsor/11/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/12/website" target="_blank"><img src="https://opencollective.com/split/sponsor/12/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/13/website" target="_blank"><img src="https://opencollective.com/split/sponsor/13/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/14/website" target="_blank"><img src="https://opencollective.com/split/sponsor/14/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/15/website" target="_blank"><img src="https://opencollective.com/split/sponsor/15/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/16/website" target="_blank"><img src="https://opencollective.com/split/sponsor/16/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/17/website" target="_blank"><img src="https://opencollective.com/split/sponsor/17/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/18/website" target="_blank"><img src="https://opencollective.com/split/sponsor/18/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/19/website" target="_blank"><img src="https://opencollective.com/split/sponsor/19/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/20/website" target="_blank"><img src="https://opencollective.com/split/sponsor/20/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/21/website" target="_blank"><img src="https://opencollective.com/split/sponsor/21/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/22/website" target="_blank"><img src="https://opencollective.com/split/sponsor/22/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/23/website" target="_blank"><img src="https://opencollective.com/split/sponsor/23/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/24/website" target="_blank"><img src="https://opencollective.com/split/sponsor/24/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/25/website" target="_blank"><img src="https://opencollective.com/split/sponsor/25/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/26/website" target="_blank"><img src="https://opencollective.com/split/sponsor/26/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/27/website" target="_blank"><img src="https://opencollective.com/split/sponsor/27/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/28/website" target="_blank"><img src="https://opencollective.com/split/sponsor/28/avatar.svg"></a>
<a href="https://opencollective.com/split/sponsor/29/website" target="_blank"><img src="https://opencollective.com/split/sponsor/29/avatar.svg"></a>
## Contribute
Please do! Over 70 different people have contributed to the project, you can see them all here: https://github.com/splitrb/split/graphs/contributors.
### Development
The source code is hosted at [GitHub](https://github.com/splitrb/split).
Report issues and feature requests on [GitHub Issues](https://github.com/splitrb/split/issues).
You can find a discussion form on [Google Groups](https://groups.google.com/d/forum/split-ruby).
### Tests
Run the tests like this:
# Start a Redis server in another tab.
redis-server
bundle
rake spec
### A Note on Patches and Pull Requests
* Fork the project.
* Make your feature addition or bug fix.
* Add tests for it. This is important so I don't break it in a
future version unintentionally.
* Add documentation if necessary.
* Commit. Do not mess with the rakefile, version, or history.
(If you want to have your own version, that is fine. But bump the version in a commit by itself, which I can ignore when I pull.)
* Send a pull request. Bonus points for topic branches.
### Code of Conduct
Please note that this project is released with a [Contributor Code of Conduct](CODE_OF_CONDUCT.md). By participating in this project you agree to abide by its terms.
## Copyright
[MIT License](LICENSE) © 2019 [Andrew Nesbitt](https://github.com/andrew).
<MSG> Merge pull request #367 from hkliya/patch-2
Update README.md
<DFF> @@ -493,7 +493,7 @@ Split.configure do |config|
config.experiments = {
my_first_experiment: {
alternatives: ["a", "b"],
- metric: :my_metric,
+ metric: :my_metric
}
}
end
| 1 | Merge pull request #367 from hkliya/patch-2 | 1 | .md | md | mit | splitrb/split |
10070235 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
it 'should delete itself' do
experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
experiment.save
experiment.delete
Split.redis.exists('basket_text').should be false
lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Deleting an experiment should also increment the version
<DFF> @@ -27,14 +27,23 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
-
- it 'should delete itself' do
- experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
- experiment.save
+
+ describe 'deleting' do
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
+ end
- experiment.delete
- Split.redis.exists('basket_text').should be false
- lambda { Split::Experiment.find('link_color') }.should raise_error
+ it "should increment the version" do
+ experiment = Split::Experiment.find_or_create('link_color', 'blue', 'red', 'green')
+ experiment.version.should eql(0)
+ experiment.delete
+ experiment.version.should eql(1)
+ end
end
describe 'new record?' do
| 16 | Deleting an experiment should also increment the version | 7 | .rb | rb | mit | splitrb/split |
10070236 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
it 'should delete itself' do
experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
experiment.save
experiment.delete
Split.redis.exists('basket_text').should be false
lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Deleting an experiment should also increment the version
<DFF> @@ -27,14 +27,23 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
-
- it 'should delete itself' do
- experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
- experiment.save
+
+ describe 'deleting' do
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
+ end
- experiment.delete
- Split.redis.exists('basket_text').should be false
- lambda { Split::Experiment.find('link_color') }.should raise_error
+ it "should increment the version" do
+ experiment = Split::Experiment.find_or_create('link_color', 'blue', 'red', 'green')
+ experiment.version.should eql(0)
+ experiment.delete
+ experiment.version.should eql(1)
+ end
end
describe 'new record?' do
| 16 | Deleting an experiment should also increment the version | 7 | .rb | rb | mit | splitrb/split |
10070237 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
it 'should delete itself' do
experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
experiment.save
experiment.delete
Split.redis.exists('basket_text').should be false
lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Deleting an experiment should also increment the version
<DFF> @@ -27,14 +27,23 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
-
- it 'should delete itself' do
- experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
- experiment.save
+
+ describe 'deleting' do
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
+ end
- experiment.delete
- Split.redis.exists('basket_text').should be false
- lambda { Split::Experiment.find('link_color') }.should raise_error
+ it "should increment the version" do
+ experiment = Split::Experiment.find_or_create('link_color', 'blue', 'red', 'green')
+ experiment.version.should eql(0)
+ experiment.delete
+ experiment.version.should eql(1)
+ end
end
describe 'new record?' do
| 16 | Deleting an experiment should also increment the version | 7 | .rb | rb | mit | splitrb/split |
10070238 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Delete an experiment
<DFF> @@ -27,6 +27,14 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
+
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
| 8 | Delete an experiment | 0 | .rb | rb | mit | splitrb/split |
10070239 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Delete an experiment
<DFF> @@ -27,6 +27,14 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
+
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
| 8 | Delete an experiment | 0 | .rb | rb | mit | splitrb/split |
10070240 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Delete an experiment
<DFF> @@ -27,6 +27,14 @@ describe Split::Experiment do
Split.redis.exists('basket_text').should be true
Split.redis.lrange('basket_text', 0, -1).should eql(['Basket', "Cart"])
end
+
+ it 'should delete itself' do
+ experiment = Split::Experiment.new('basket_text', 'Basket', "Cart")
+ experiment.save
+
+ experiment.delete
+ Split.redis.exists('basket_text').should be false
+ end
describe 'new record?' do
it "should know if it hasn't been saved yet" do
| 8 | Delete an experiment | 0 | .rb | rb | mit | splitrb/split |
10070241 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
experiment.delete
Split.redis.exists('basket_text').should be false
end
describe 'new record?' do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Refactored deleting specs a little
<DFF> @@ -34,6 +34,7 @@ describe Split::Experiment do
experiment.delete
Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
| 1 | Refactored deleting specs a little | 0 | .rb | rb | mit | splitrb/split |
10070242 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
experiment.delete
Split.redis.exists('basket_text').should be false
end
describe 'new record?' do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Refactored deleting specs a little
<DFF> @@ -34,6 +34,7 @@ describe Split::Experiment do
experiment.delete
Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
| 1 | Refactored deleting specs a little | 0 | .rb | rb | mit | splitrb/split |
10070243 | <NME> experiment_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
require "time"
describe Split::Experiment do
def new_experiment(goals = [])
Split::Experiment.new("link_color", alternatives: ["blue", "red", "green"], goals: goals)
end
def alternative(color)
Split::Alternative.new(color, "link_color")
end
let(:experiment) { new_experiment }
let(:blue) { alternative("blue") }
let(:green) { alternative("green") }
context "with an experiment" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"]) }
it "should have a name" do
expect(experiment.name).to eq("basket_text")
end
it "should have alternatives" do
expect(experiment.alternatives.length).to be 2
end
it "should have alternatives with correct names" do
expect(experiment.alternatives.collect { |a| a.name }).to eq(["Basket", "Cart"])
end
experiment.delete
Split.redis.exists('basket_text').should be false
end
describe 'new record?' do
expect(Split.redis.exists?("basket_text")).to be true
end
it "should save the start time to redis" do
experiment_start_time = Time.at(1372167761)
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should not save the start time to redis when start_manually is enabled" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should save the selected algorithm to redis" do
experiment_algorithm = Split::Algorithms::Whiplash
experiment.algorithm = experiment_algorithm
experiment.save
expect(Split::ExperimentCatalog.find("basket_text").algorithm).to eq(experiment_algorithm)
end
it "should handle having a start time stored as a string" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).twice.and_return(experiment_start_time)
experiment.save
Split.redis.hset(:experiment_start_times, experiment.name, experiment_start_time.to_s)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to eq(experiment_start_time)
end
it "should handle not having a start time" do
experiment_start_time = Time.parse("Sat Mar 03 14:01:03")
expect(Time).to receive(:now).and_return(experiment_start_time)
experiment.save
Split.redis.hdel(:experiment_start_times, experiment.name)
expect(Split::ExperimentCatalog.find("basket_text").start_time).to be_nil
end
it "should not create duplicates when saving multiple times" do
experiment.save
experiment.save
expect(Split.redis.exists?("basket_text")).to be true
expect(Split.redis.lrange("basket_text", 0, -1)).to eq(['{"Basket":1}', '{"Cart":1}'])
end
describe "new record?" do
it "should know if it hasn't been saved yet" do
expect(experiment.new_record?).to be_truthy
end
it "should know if it has been saved yet" do
experiment.save
expect(experiment.new_record?).to be_falsey
end
end
describe "control" do
it "should be the first alternative" do
experiment.save
expect(experiment.control.name).to eq("Basket")
end
end
end
describe "initialization" do
it "should set the algorithm when passed as an option to the initializer" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should be possible to make an experiment not resettable" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
expect(experiment.resettable).to be_falsey
end
context "from configuration" do
let(:experiment_name) { :my_experiment }
let(:experiments) do
{
experiment_name => {
alternatives: ["Control Opt", "Alt one"]
}
}
end
before { Split.configuration.experiments = experiments }
it "assigns default values to the experiment" do
expect(Split::Experiment.new(experiment_name).resettable).to eq(true)
end
end
end
describe "persistent configuration" do
it "should persist resettable in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], resettable: false)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.resettable).to be_falsey
end
describe "#metadata" do
let(:experiment) { Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash, metadata: meta) }
let(:meta) { { a: "b" } }
before do
experiment.save
end
it "should delete the key when metadata is removed" do
experiment.metadata = nil
experiment.save
expect(Split.redis.exists?(experiment.metadata_key)).to be_falsey
end
context "simple hash" do
let(:meta) { { "basket" => "a", "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
context "nested hash" do
let(:meta) { { "basket" => { "one" => "two" }, "cart" => "b" } }
it "should persist metadata in redis" do
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.metadata).to eq(meta)
end
end
end
it "should persist algorithm in redis" do
experiment = Split::Experiment.new("basket_text", alternatives: ["Basket", "Cart"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("basket_text")
expect(e).to eq(experiment)
expect(e.algorithm).to eq(Split::Algorithms::Whiplash)
end
it "should persist a new experiment in redis, that does not exist in the configuration file" do
experiment = Split::Experiment.new("foobar", alternatives: ["tra", "la"], algorithm: Split::Algorithms::Whiplash)
experiment.save
e = Split::ExperimentCatalog.find("foobar")
expect(e).to eq(experiment)
expect(e.alternatives.collect { |a| a.name }).to eq(["tra", "la"])
end
end
describe "deleting" do
it "should delete itself" do
experiment = Split::Experiment.new("basket_text", alternatives: [ "Basket", "Cart"])
experiment.save
experiment.delete
expect(Split.redis.exists?("link_color")).to be false
expect(Split::ExperimentCatalog.find("link_color")).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.delete
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_delete hook" do
expect(Split.configuration.on_experiment_delete).to receive(:call)
experiment.delete
end
it "should call the on_before_experiment_delete hook" do
expect(Split.configuration.on_before_experiment_delete).to receive(:call)
experiment.delete
end
it "should reset the start time if the experiment should be manually started" do
Split.configuration.start_manually = true
experiment.start
experiment.delete
expect(experiment.start_time).to be_nil
end
it "should default cohorting back to false" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq(true)
experiment.delete
expect(experiment.cohorting_disabled?).to eq(false)
end
end
describe "winner" do
it "should have no winner initially" do
expect(experiment.winner).to be_nil
end
end
describe "winner=" do
it "should allow you to specify a winner" do
experiment.save
experiment.winner = "red"
expect(experiment.winner.name).to eq("red")
end
it "should call the on_experiment_winner_choose hook" do
expect(Split.configuration.on_experiment_winner_choose).to receive(:call)
experiment.winner = "green"
end
context "when has_winner state is memoized" do
before { expect(experiment).to_not have_winner }
it "should keep has_winner state consistent" do
experiment.winner = "red"
expect(experiment).to have_winner
end
end
end
describe "reset_winner" do
before { experiment.winner = "green" }
it "should reset the winner" do
experiment.reset_winner
expect(experiment.winner).to be_nil
end
context "when has_winner state is memoized" do
before { expect(experiment).to have_winner }
it "should keep has_winner state consistent" do
experiment.reset_winner
expect(experiment).to_not have_winner
end
end
end
describe "has_winner?" do
context "with winner" do
before { experiment.winner = "red" }
it "returns true" do
expect(experiment).to have_winner
end
end
context "without winner" do
it "returns false" do
expect(experiment).to_not have_winner
end
end
it "memoizes has_winner state" do
expect(experiment).to receive(:winner).once
expect(experiment).to_not have_winner
expect(experiment).to_not have_winner
end
end
describe "reset" do
let(:reset_manually) { false }
before do
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
experiment.save
green.increment_participation
green.increment_participation
end
it "should reset all alternatives" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
it "should reset the winner" do
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
experiment.reset
expect(experiment.winner).to be_nil
end
it "should increment the version" do
expect(experiment.version).to eq(0)
experiment.reset
expect(experiment.version).to eq(1)
end
it "should call the on_experiment_reset hook" do
expect(Split.configuration.on_experiment_reset).to receive(:call)
experiment.reset
end
it "should call the on_before_experiment_reset hook" do
expect(Split.configuration.on_before_experiment_reset).to receive(:call)
experiment.reset
end
end
describe "algorithm" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
it "should use the default algorithm if none is specified" do
expect(experiment.algorithm).to eq(Split.configuration.algorithm)
end
it "should use the user specified algorithm for this experiment if specified" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to eq(Split::Algorithms::Whiplash)
end
end
describe "#next_alternative" do
context "with multiple alternatives" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue", "red", "green") }
context "with winner" do
it "should always return the winner" do
green = Split::Alternative.new("green", "link_color")
experiment.winner = "green"
expect(experiment.next_alternative.name).to eq("green")
green.increment_participation
expect(experiment.next_alternative.name).to eq("green")
end
end
context "without winner" do
it "should use the specified algorithm" do
experiment.algorithm = Split::Algorithms::Whiplash
expect(experiment.algorithm).to receive(:choose_alternative).and_return(Split::Alternative.new("green", "link_color"))
expect(experiment.next_alternative.name).to eq("green")
end
end
end
context "with single alternative" do
let(:experiment) { Split::ExperimentCatalog.find_or_create("link_color", "blue") }
it "should always return the only alternative" do
expect(experiment.next_alternative.name).to eq("blue")
expect(experiment.next_alternative.name).to eq("blue")
end
end
end
describe "#cohorting_disabled?" do
it "returns false when nothing has been configured" do
expect(experiment.cohorting_disabled?).to eq false
end
it "returns true when enable_cohorting is performed" do
experiment.enable_cohorting
expect(experiment.cohorting_disabled?).to eq false
end
it "returns false when nothing has been configured" do
experiment.disable_cohorting
expect(experiment.cohorting_disabled?).to eq true
end
end
describe "changing an existing experiment" do
def same_but_different_alternative
Split::ExperimentCatalog.find_or_create("link_color", "blue", "yellow", "orange")
end
it "should reset an experiment if it is loaded with different alternatives" do
experiment.save
blue.participant_count = 5
same_experiment = same_but_different_alternative
expect(same_experiment.alternatives.map(&:name)).to eq(["blue", "yellow", "orange"])
expect(blue.participant_count).to eq(0)
end
it "should only reset once" do
experiment.save
expect(experiment.version).to eq(0)
same_experiment = same_but_different_alternative
expect(same_experiment.version).to eq(1)
same_experiment_again = same_but_different_alternative
expect(same_experiment_again.version).to eq(1)
end
context "when metadata is changed" do
it "should increase version" do
experiment.save
experiment.metadata = { "foo" => "bar" }
expect { experiment.save }.to change { experiment.version }.by(1)
end
it "does not increase version" do
experiment.metadata = nil
experiment.save
expect { experiment.save }.to change { experiment.version }.by(0)
end
end
context "when experiment configuration is changed" do
let(:reset_manually) { false }
before do
experiment.save
allow(Split.configuration).to receive(:reset_manually).and_return(reset_manually)
green.increment_participation
green.increment_participation
experiment.set_alternatives_and_options(alternatives: %w(blue red green zip),
goals: %w(purchase))
experiment.save
end
it "resets all alternatives" do
expect(green.participant_count).to eq(0)
expect(green.completed_count).to eq(0)
end
context "when reset_manually is set" do
let(:reset_manually) { true }
it "does not reset alternatives" do
expect(green.participant_count).to eq(2)
expect(green.completed_count).to eq(0)
end
end
end
end
describe "alternatives passed as non-strings" do
it "should throw an exception if an alternative is passed that is not a string" do
expect { Split::ExperimentCatalog.find_or_create("link_color", :blue, :red) }.to raise_error(ArgumentError)
expect { Split::ExperimentCatalog.find_or_create("link_enabled", true, false) }.to raise_error(ArgumentError)
end
end
describe "specifying weights" do
let(:experiment_with_weight) {
Split::ExperimentCatalog.find_or_create("link_color", { "blue" => 1 }, { "red" => 2 })
}
it "should work for a new experiment" do
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
it "should work for an existing experiment" do
experiment.save
expect(experiment_with_weight.alternatives.map(&:weight)).to eq([1, 2])
end
end
describe "specifying goals" do
let(:experiment) {
new_experiment(["purchase"])
}
context "saving experiment" do
let(:same_but_different_goals) { Split::ExperimentCatalog.find_or_create({ "link_color" => ["purchase", "refund"] }, "blue", "red", "green") }
before { experiment.save }
it "can find existing experiment" do
expect(Split::ExperimentCatalog.find("link_color").name).to eq("link_color")
end
it "should reset an experiment if it is loaded with different goals" do
same_but_different_goals
expect(Split::ExperimentCatalog.find("link_color").goals).to eq(["purchase", "refund"])
end
end
it "should have goals" do
expect(experiment.goals).to eq(["purchase"])
end
context "find or create experiment" do
it "should have correct goals" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.goals).to eq(["purchase", "refund"])
experiment = Split::ExperimentCatalog.find_or_create("link_color3", "blue", "red", "green")
expect(experiment.goals).to eq([])
end
end
end
describe "beta probability calculation" do
it "should return a hash with the probability of each alternative being the best" do
experiment = Split::ExperimentCatalog.find_or_create("mathematicians", "bernoulli", "poisson", "lagrange")
experiment.calc_winning_alternatives
expect(experiment.alternative_probabilities).not_to be_nil
end
it "should return between 46% and 54% probability for an experiment with 2 alternatives and no data" do
experiment = Split::ExperimentCatalog.find_or_create("scientists", "einstein", "bohr")
experiment.calc_winning_alternatives
expect(experiment.alternatives[0].p_winner).to be_within(0.04).of(0.50)
end
it "should calculate the probability of being the winning alternative separately for each goal", skip: true do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
goal1 = experiment.goals[0]
goal2 = experiment.goals[1]
experiment.alternatives.each do |alternative|
alternative.participant_count = 50
alternative.set_completed_count(10, goal1)
alternative.set_completed_count(15+rand(30), goal2)
end
experiment.calc_winning_alternatives
alt = experiment.alternatives[0]
p_goal1 = alt.p_winner(goal1)
p_goal2 = alt.p_winner(goal2)
expect(p_goal1).not_to be_within(0.04).of(p_goal2)
end
it "should return nil and not re-calculate probabilities if they have already been calculated today" do
experiment = Split::ExperimentCatalog.find_or_create({ "link_color3" => ["purchase", "refund"] }, "blue", "red", "green")
expect(experiment.calc_winning_alternatives).not_to be nil
expect(experiment.calc_winning_alternatives).to be nil
end
end
end
<MSG> Refactored deleting specs a little
<DFF> @@ -34,6 +34,7 @@ describe Split::Experiment do
experiment.delete
Split.redis.exists('basket_text').should be false
+ lambda { Split::Experiment.find('link_color') }.should raise_error
end
describe 'new record?' do
| 1 | Refactored deleting specs a little | 0 | .rb | rb | mit | splitrb/split |
10070244 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
small.participant_count.should eql(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Written failing test for incorrectly raising the completion count
<DFF> @@ -103,6 +103,23 @@ describe Split::Helper do
small.participant_count.should eql(0)
end
+ it 'should not raise the completion rate of an experiment that the user is not participating in' do
+ ab_test('link_color', 'blue', 'red')
+ ab_test('button_size', 'small', 'big')
+
+ # So, user should be participating in the link_color experiment and
+ # receive the control for button_size. As the user is not participating in
+ # the button size experiment, finishing it should not increase the
+ # completion count for that alternative.
+ ab_user['button_size'].should eql('small')
+ previous_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+
+ finished('button_size')
+
+ new_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+ previous_completion_count.should eql(new_completion_count)
+ end
+
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
| 17 | Written failing test for incorrectly raising the completion count | 0 | .rb | rb | mit | splitrb/split |
10070245 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
small.participant_count.should eql(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Written failing test for incorrectly raising the completion count
<DFF> @@ -103,6 +103,23 @@ describe Split::Helper do
small.participant_count.should eql(0)
end
+ it 'should not raise the completion rate of an experiment that the user is not participating in' do
+ ab_test('link_color', 'blue', 'red')
+ ab_test('button_size', 'small', 'big')
+
+ # So, user should be participating in the link_color experiment and
+ # receive the control for button_size. As the user is not participating in
+ # the button size experiment, finishing it should not increase the
+ # completion count for that alternative.
+ ab_user['button_size'].should eql('small')
+ previous_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+
+ finished('button_size')
+
+ new_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+ previous_completion_count.should eql(new_completion_count)
+ end
+
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
| 17 | Written failing test for incorrectly raising the completion count | 0 | .rb | rb | mit | splitrb/split |
10070246 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
small.participant_count.should eql(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
end
it "should not change the user state when reset is true" do
expect(self).not_to receive(:reset!)
ab_finished(:not_started_experiment)
end
it "should not increment the completed counter" do
ab_finished(:not_started_experiment)
expect(Split::Alternative.new("control", :not_started_experiment).completed_count).to eq(0)
expect(Split::Alternative.new("alt", :not_started_experiment).completed_count).to eq(0)
end
end
end
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Written failing test for incorrectly raising the completion count
<DFF> @@ -103,6 +103,23 @@ describe Split::Helper do
small.participant_count.should eql(0)
end
+ it 'should not raise the completion rate of an experiment that the user is not participating in' do
+ ab_test('link_color', 'blue', 'red')
+ ab_test('button_size', 'small', 'big')
+
+ # So, user should be participating in the link_color experiment and
+ # receive the control for button_size. As the user is not participating in
+ # the button size experiment, finishing it should not increase the
+ # completion count for that alternative.
+ ab_user['button_size'].should eql('small')
+ previous_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+
+ finished('button_size')
+
+ new_completion_count = Split::Alternative.new('small', 'button_size').completed_count
+ previous_completion_count.should eql(new_completion_count)
+ end
+
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
| 17 | Written failing test for incorrectly raising the completion count | 0 | .rb | rb | mit | splitrb/split |
10070247 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it 'should show multiple tests' do
Split.configure do |config|
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning
fix active experiments when experiment is on a later version
<DFF> @@ -449,6 +449,14 @@ describe Split::Helper do
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
+
+ it 'should show an active test when an experiment is on a later version' do
+ experiment.reset
+ expect(experiment.version).to eq(1)
+ ab_test('link_color', 'blue', 'red')
+ expect(active_experiments.count).to eq 1
+ expect(active_experiments.first[0]).to eq "link_color"
+ end
it 'should show multiple tests' do
Split.configure do |config|
| 8 | Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning | 0 | .rb | rb | mit | splitrb/split |
10070248 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it 'should show multiple tests' do
Split.configure do |config|
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning
fix active experiments when experiment is on a later version
<DFF> @@ -449,6 +449,14 @@ describe Split::Helper do
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
+
+ it 'should show an active test when an experiment is on a later version' do
+ experiment.reset
+ expect(experiment.version).to eq(1)
+ ab_test('link_color', 'blue', 'red')
+ expect(active_experiments.count).to eq 1
+ expect(active_experiments.first[0]).to eq "link_color"
+ end
it 'should show multiple tests' do
Split.configure do |config|
| 8 | Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning | 0 | .rb | rb | mit | splitrb/split |
10070249 | <NME> helper_spec.rb
<BEF> # frozen_string_literal: true
require "spec_helper"
# TODO change some of these tests to use Rack::Test
describe Split::Helper do
include Split::Helper
let(:experiment) {
Split::ExperimentCatalog.find_or_create("link_color", "blue", "red")
}
describe "ab_test" do
it "should not raise an error when passed strings for alternatives" do
expect { ab_test("xyz", "1", "2", "3") }.not_to raise_error
end
it "should not raise an error when passed an array for alternatives" do
expect { ab_test("xyz", ["1", "2", "3"]) }.not_to raise_error
end
it "should raise the appropriate error when passed integers for alternatives" do
expect { ab_test("xyz", 1, 2, 3) }.to raise_error(ArgumentError)
end
it "should raise the appropriate error when passed symbols for alternatives" do
expect { ab_test("xyz", :a, :b, :c) }.to raise_error(ArgumentError)
end
it "should not raise error when passed an array for goals" do
expect { ab_test({ "link_color" => ["purchase", "refund"] }, "blue", "red") }.not_to raise_error
end
it "should not raise error when passed just one goal" do
expect { ab_test({ "link_color" => "purchase" }, "blue", "red") }.not_to raise_error
end
it "raises an appropriate error when processing combined expirements" do
Split.configuration.experiments = {
combined_exp_1: {
alternatives: [ { name: "control", percent: 50 }, { name: "test-alt", percent: 50 } ],
metric: :my_metric,
combined_experiments: [:combined_exp_1_sub_1]
}
}
Split::ExperimentCatalog.find_or_create("combined_exp_1")
expect { ab_test("combined_exp_1") }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "should assign a random alternative to a new user when there are an equal number of alternatives assigned" do
ab_test("link_color", "blue", "red")
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should increment the participation counter after assignment to a new user" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count + 1)
end
it "should not increment the counter for an experiment that the user is not participating in" do
ab_test("link_color", "blue", "red")
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
# User shouldn't participate in this second experiment
ab_test("button_size", "small", "big")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should not increment the counter for an not started experiment" do
expect(Split.configuration).to receive(:start_manually).and_return(true)
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
expect {
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
}.not_to change { e.participant_count }
end
it "should return the given alternative for an existing user" do
expect(ab_test("link_color", "blue", "red")).to eq ab_test("link_color", "blue", "red")
end
it "should always return the winner if one is present" do
experiment.winner = "orange"
expect(ab_test("link_color", "blue", "red")).to eq("orange")
end
it "should allow the alternative to be forced by passing it in the params" do
# ?ab_test[link_color]=blue
@params = { "ab_test" => { "link_color" => "blue" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
@params = { "ab_test" => { "link_color" => "red" } }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "blue" => 5 }, "red" => 1)
expect(alternative).to eq("red")
end
it "should not allow an arbitrary alternative" do
@params = { "ab_test" => { "link_color" => "pink" } }
alternative = ab_test("link_color", "blue")
expect(alternative).to eq("blue")
end
it "should not store the split when a param forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
it "SPLIT_DISABLE query parameter should also force the alternative (uses control)" do
@params = { "SPLIT_DISABLE" => "true" }
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq("blue")
alternative = ab_test("link_color", { "blue" => 1 }, "red" => 5)
expect(alternative).to eq("blue")
alternative = ab_test("link_color", "red", "blue")
expect(alternative).to eq("red")
alternative = ab_test("link_color", { "red" => 5 }, "blue" => 1)
expect(alternative).to eq("red")
end
it "should not store the split when Split generically disabled" do
@params = { "SPLIT_DISABLE" => "true" }
expect(ab_user).not_to receive(:[]=)
ab_test("link_color", "blue", "red")
end
context "when store_override is set" do
before { Split.configuration.store_override = true }
it "should store the forced alternative" do
@params = { "ab_test" => { "link_color" => "blue" } }
expect(ab_user).to receive(:[]=).with("link_color", "blue")
ab_test("link_color", "blue", "red")
end
end
context "when on_trial_choose is set" do
before { Split.configuration.on_trial_choose = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_test("link_color", "blue", "red")
end
end
it "should allow passing a block" do
alt = ab_test("link_color", "blue", "red")
ret = ab_test("link_color", "blue", "red") { |alternative| "shared/#{alternative}" }
expect(ret).to eq("shared/#{alt}")
end
it "should allow the share of visitors see an alternative to be specified" do
ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })
expect(["red", "blue"]).to include(ab_user["link_color"])
end
it "should allow alternative weighting interface as a single hash" do
ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.alternatives.map(&:name)).to eq(["blue", "red"])
expect(experiment.alternatives.collect { |a| a.weight }).to match_array([0.01, 0.2])
end
it "should only let a user participate in one experiment at a time" do
link_color = ab_test("link_color", "blue", "red")
ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
big = Split::Alternative.new("big", "button_size")
expect(big.participant_count).to eq(0)
small = Split::Alternative.new("small", "button_size")
expect(small.participant_count).to eq(0)
end
it "should let a user participate in many experiment with allow_multiple_experiments option" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
link_color = ab_test("link_color", "blue", "red")
button_size = ab_test("button_size", "small", "big")
expect(ab_user["link_color"]).to eq(link_color)
expect(ab_user["button_size"]).to eq(button_size)
button_size_alt = Split::Alternative.new(button_size, "button_size")
expect(button_size_alt.participant_count).to eq(1)
end
context "with allow_multiple_experiments = 'control'" do
it "should let a user participate in many experiment with one non-'control' alternative" do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
groups = 100.times.map do |n|
ab_test("test#{n}".to_sym, { "control" => (100 - n) }, { "test#{n}-alt" => n })
end
experiments = ab_user.active_experiments
expect(experiments.size).to be > 1
count_control = experiments.values.count { |g| g == "control" }
expect(count_control).to eq(experiments.size - 1)
count_alts = groups.count { |g| g != "control" }
expect(count_alts).to eq(1)
end
context "when user already has experiment" do
let(:mock_user) { Split::User.new(self, { "test_0" => "test-alt" }) }
before do
Split.configure do |config|
config.allow_multiple_experiments = "control"
end
Split::ExperimentCatalog.find_or_initialize("test_0", "control", "test-alt").save
Split::ExperimentCatalog.find_or_initialize("test_1", "control", "test-alt").save
end
it "should restore previously selected alternative" do
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 1 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "should select the correct alternatives after experiment resets" do
experiment = Split::ExperimentCatalog.find(:test_0)
experiment.reset
mock_user[experiment.key] = "test-alt"
expect(ab_user.active_experiments.size).to eq 1
expect(ab_test(:test_0, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "test-alt"
end
it "lets override existing choice" do
pending "this requires user store reset on first call not depending on whelther it is current trial"
@params = { "ab_test" => { "test_1" => "test-alt" } }
expect(ab_test(:test_0, { "control" => 0 }, { "test-alt" => 100 })).to eq "control"
expect(ab_test(:test_1, { "control" => 100 }, { "test-alt" => 1 })).to eq "test-alt"
end
end
end
it "should not over-write a finished key when an experiment is on a later version" do
experiment.increment_version
ab_user = { experiment.key => "blue", experiment.finished_key => true }
finished_session = ab_user.dup
ab_test("link_color", "blue", "red")
expect(ab_user).to eq(finished_session)
end
end
describe "metadata" do
context "is defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: { "one" => "Meta1", "two" => "Meta2" }
}
}
end
it "should be passed to helper block" do
@params = { "ab_test" => { "my_experiment" => "two" } }
expect(ab_test("my_experiment")).to eq "two"
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq("Meta2")
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment")).to eq "one"
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq("Meta1")
end
end
context "is not defined" do
before do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
metadata: nil
}
}
end
it "should be passed to helper block" do
expect(ab_test("my_experiment") do |alternative, meta|
meta
end).to eq({})
end
it "should pass control metadata helper block if library disabled" do
Split.configure do |config|
config.enabled = false
end
expect(ab_test("my_experiment") do |_, meta|
meta
end).to eq({})
end
end
end
describe "ab_finished" do
context "for an experiment that the user participates in" do
before(:each) do
@experiment_name = "link_color"
@alternatives = ["blue", "red"]
@experiment = Split::ExperimentCatalog.find_or_create(@experiment_name, *@alternatives)
@alternative_name = ab_test(@experiment_name, *@alternatives)
@previous_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
end
it "should increment the counter for the completed alternative" do
ab_finished(@experiment_name)
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should set experiment's finished key if reset is false" do
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should not increment the counter if reset is false and the experiment has been already finished" do
2.times { ab_finished(@experiment_name, { reset: false }) }
new_completion_count = Split::Alternative.new(@alternative_name, @experiment_name).completed_count
expect(new_completion_count).to eq(@previous_completion_count + 1)
end
it "should not increment the counter for an ended experiment" do
e = Split::ExperimentCatalog.find_or_create("button_size", "small", "big")
e.winner = "small"
a = ab_test("button_size", "small", "big")
expect(a).to eq("small")
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new(a, "button_size").completed_count }
end
it "should clear out the user's participation from their session" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should not clear out the users session if reset is false" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name, { reset: false })
expect(ab_user[@experiment.key]).to eq(@alternative_name)
expect(ab_user[@experiment.finished_key]).to eq(true)
end
it "should reset the users session when experiment is not versioned" do
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
it "should reset the users session when experiment is versioned" do
@experiment.increment_version
@alternative_name = ab_test(@experiment_name, *@alternatives)
expect(ab_user[@experiment.key]).to eq(@alternative_name)
ab_finished(@experiment_name)
expect(ab_user.keys).to be_empty
end
context "when on_trial_complete is set" do
before { Split.configuration.on_trial_complete = :some_method }
it "should call the method" do
expect(self).to receive(:some_method)
ab_finished(@experiment_name)
end
it "should not call the method without alternative" do
ab_user[@experiment.key] = nil
expect(self).not_to receive(:some_method)
ab_finished(@experiment_name)
end
end
end
context "for an experiment that the user is excluded from" do
before do
alternative = ab_test("link_color", "blue", "red")
expect(Split::Alternative.new(alternative, "link_color").participant_count).to eq(1)
alternative = ab_test("button_size", "small", "big")
expect(Split::Alternative.new(alternative, "button_size").participant_count).to eq(0)
end
it "should not increment the completed counter" do
# So, user should be participating in the link_color experiment and
# receive the control for button_size. As the user is not participating in
# the button size experiment, finishing it should not increase the
# completion count for that alternative.
expect {
ab_finished("button_size")
}.not_to change { Split::Alternative.new("small", "button_size").completed_count }
end
end
context "for an experiment that the user does not participate in" do
before do
Split::ExperimentCatalog.find_or_create(:not_started_experiment, "control", "alt")
end
it "should not raise an exception" do
expect { ab_finished(:not_started_experiment) }.not_to raise_exception
end
it "should not change the user state when reset is false" do
expect { ab_finished(:not_started_experiment, reset: false) }.not_to change { ab_user.keys }.from([])
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it 'should show multiple tests' do
Split.configure do |config|
context "finished with config" do
it "passes reset option" do
Split.configuration.experiments = {
my_experiment: {
alternatives: ["one", "two"],
resettable: false,
}
}
alternative = ab_test(:my_experiment)
experiment = Split::ExperimentCatalog.find :my_experiment
ab_finished :my_experiment
expect(ab_user[experiment.key]).to eq(alternative)
expect(ab_user[experiment.finished_key]).to eq(true)
end
end
context "finished with metric name" do
before { Split.configuration.experiments = {} }
before { expect(Split::Alternative).to receive(:new).at_least(1).times.and_call_original }
def should_finish_experiment(experiment_name, should_finish = true)
alts = Split.configuration.experiments[experiment_name][:alternatives]
experiment = Split::ExperimentCatalog.find_or_create(experiment_name, *alts)
alt_name = ab_user[experiment.key] = alts.first
alt = double("alternative")
expect(alt).to receive(:name).at_most(1).times.and_return(alt_name)
expect(Split::Alternative).to receive(:new).at_most(1).times.with(alt_name, experiment_name.to_s).and_return(alt)
if should_finish
expect(alt).to receive(:increment_completion).at_most(1).times
else
expect(alt).not_to receive(:increment_completion)
end
end
it "completes the test" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
metric: :my_metric
}
should_finish_experiment :my_experiment
ab_finished :my_metric
end
it "completes all relevant tests" do
Split.configuration.experiments = {
exp_1: {
alternatives: [ "1-1", "1-2" ],
metric: :my_metric
},
exp_2: {
alternatives: [ "2-1", "2-2" ],
metric: :another_metric
},
exp_3: {
alternatives: [ "3-1", "3-2" ],
metric: :my_metric
},
}
should_finish_experiment :exp_1
should_finish_experiment :exp_2, false
should_finish_experiment :exp_3
ab_finished :my_metric
end
it "passes reset option" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
resettable: false,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
it "passes through options" do
Split.configuration.experiments = {
my_exp: {
alternatives: ["one", "two"],
metric: :my_metric,
}
}
alternative_name = ab_test(:my_exp)
exp = Split::ExperimentCatalog.find :my_exp
ab_finished :my_metric, reset: false
expect(ab_user[exp.key]).to eq(alternative_name)
expect(ab_user[exp.finished_key]).to be_truthy
end
end
describe "conversions" do
it "should return a conversion rate for an alternative" do
alternative_name = ab_test("link_color", "blue", "red")
previous_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(previous_convertion_rate).to eq(0.0)
ab_finished("link_color")
new_convertion_rate = Split::Alternative.new(alternative_name, "link_color").conversion_rate
expect(new_convertion_rate).to eq(1.0)
end
end
describe "active experiments" do
it "should show an active test" do
alternative = ab_test("def", "4", "5", "6")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show a finished test" do
alternative = ab_test("def", "4", "5", "6")
ab_finished("def", { reset: false })
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
it "should show an active test when an experiment is on a later version" do
experiment.reset
expect(experiment.version).to eq(1)
ab_test("link_color", "blue", "red")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "link_color"
end
it "should show versioned tests properly" do
10.times { experiment.reset }
alternative = ab_test(experiment.name, "blue", "red")
ab_finished(experiment.name, reset: false)
expect(experiment.version).to eq(10)
expect(active_experiments.count).to eq 1
expect(active_experiments).to eq({ "link_color" => alternative })
end
it "should show multiple tests" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
alternative = ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 2
expect(active_experiments["def"]).to eq alternative
expect(active_experiments["ghi"]).to eq another_alternative
end
it "should not show tests with winners" do
Split.configure do |config|
config.allow_multiple_experiments = true
end
e = Split::ExperimentCatalog.find_or_create("def", "4", "5", "6")
e.winner = "4"
ab_test("def", "4", "5", "6")
another_alternative = ab_test("ghi", "7", "8", "9")
expect(active_experiments.count).to eq 1
expect(active_experiments.first[0]).to eq "ghi"
expect(active_experiments.first[1]).to eq another_alternative
end
end
describe "when user is a robot" do
before(:each) do
@request = OpenStruct.new(user_agent: "Googlebot/2.1 (+http://www.google.com/bot.html)")
end
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not create a experiment" do
ab_test("link_color", "blue", "red")
expect(Split::Experiment.new("link_color")).to be_a_new_record
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when providing custom ignore logic" do
context "using a proc to configure custom logic" do
before(:each) do
Split.configure do |c|
c.ignore_filter = proc { |request| true } # ignore everything
end
end
it "ignores the ab_test" do
ab_test("link_color", "blue", "red")
red_count = Split::Alternative.new("red", "link_color").participant_count
blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((red_count + blue_count)).to be(0)
end
end
end
shared_examples_for "a disabled test" do
describe "ab_test" do
it "should return the control" do
alternative = ab_test("link_color", "blue", "red")
expect(alternative).to eq experiment.control.name
end
it "should not increment the participation count" do
previous_red_count = Split::Alternative.new("red", "link_color").participant_count
previous_blue_count = Split::Alternative.new("blue", "link_color").participant_count
ab_test("link_color", "blue", "red")
new_red_count = Split::Alternative.new("red", "link_color").participant_count
new_blue_count = Split::Alternative.new("blue", "link_color").participant_count
expect((new_red_count + new_blue_count)).to eq(previous_red_count + previous_blue_count)
end
end
describe "finished" do
it "should not increment the completed count" do
alternative_name = ab_test("link_color", "blue", "red")
previous_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
ab_finished("link_color")
new_completion_count = Split::Alternative.new(alternative_name, "link_color").completed_count
expect(new_completion_count).to eq(previous_completion_count)
end
end
end
describe "when ip address is ignored" do
context "individually" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.130")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it_behaves_like "a disabled test"
end
context "for a range" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.129")
Split.configure do |c|
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "using both a range and a specific value" do
before(:each) do
@request = OpenStruct.new(ip: "81.19.48.128")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
c.ignore_ip_addresses << /81\.19\.48\.[0-9]+/
end
end
it_behaves_like "a disabled test"
end
context "when ignored other address" do
before do
@request = OpenStruct.new(ip: "1.1.1.1")
Split.configure do |c|
c.ignore_ip_addresses << "81.19.48.130"
end
end
it "works as usual" do
alternative_name = ab_test("link_color", "red", "blue")
expect {
ab_finished("link_color")
}.to change(Split::Alternative.new(alternative_name, "link_color"), :completed_count).by(1)
end
end
end
describe "when user is previewing" do
before(:each) do
@request = OpenStruct.new(headers: { "x-purpose" => "preview" })
end
it_behaves_like "a disabled test"
end
describe "versioned experiments" do
it "should use version zero if no version is present" do
alternative_name = ab_test("link_color", "blue", "red")
expect(experiment.version).to eq(0)
expect(ab_user["link_color"]).to eq(alternative_name)
end
it "should save the version of the experiment to the session" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
end
it "should load the experiment even if the version is not 0" do
experiment.reset
expect(experiment.version).to eq(1)
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(alternative_name)
return_alternative_name = ab_test("link_color", "blue", "red")
expect(return_alternative_name).to eq(alternative_name)
end
it "should reset the session of a user on an older version of the experiment" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
new_alternative = Split::Alternative.new(new_alternative_name, "link_color")
expect(new_alternative.participant_count).to eq(1)
end
it "should cleanup old versions of experiments from the session" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(1)
experiment.reset
expect(experiment.version).to eq(1)
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.participant_count).to eq(0)
new_alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color:1"]).to eq(new_alternative_name)
end
it "should only count completion of users on the current version" do
alternative_name = ab_test("link_color", "blue", "red")
expect(ab_user["link_color"]).to eq(alternative_name)
Split::Alternative.new(alternative_name, "link_color")
experiment.reset
expect(experiment.version).to eq(1)
ab_finished("link_color")
alternative = Split::Alternative.new(alternative_name, "link_color")
expect(alternative.completed_count).to eq(0)
end
end
context "when redis is not available" do
before(:each) do
expect(Split).to receive(:redis).at_most(5).times.and_raise(Errno::ECONNREFUSED.new)
end
context "and db_failover config option is turned off" do
before(:each) do
Split.configure do |config|
config.db_failover = false
end
end
describe "ab_test" do
it "should raise an exception" do
expect { ab_test("link_color", "blue", "red") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "finished" do
it "should raise an exception" do
expect { ab_finished("link_color") }.to raise_error(Errno::ECONNREFUSED)
end
end
describe "disable split testing" do
before(:each) do
Split.configure do |config|
config.enabled = false
end
end
it "should not attempt to connect to redis" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should return control variable" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect { ab_finished("link_color") }.not_to raise_error
end
end
end
context "and db_failover config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover = true
end
end
describe "ab_test" do
it "should not raise an exception" do
expect { ab_test("link_color", "blue", "red") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_test("link_color", "blue", "red")
end
it "should always use first alternative" do
expect(ab_test("link_color", "blue", "red")).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("blue")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("blue")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/blue")
end
context "and db_failover_allow_parameter_override config option is turned on" do
before(:each) do
Split.configure do |config|
config.db_failover_allow_parameter_override = true
end
end
context "and given an override parameter" do
it "should use given override instead of the first alternative" do
@params = { "ab_test" => { "link_color" => "red" } }
expect(ab_test("link_color", "blue", "red")).to eq("red")
expect(ab_test("link_color", "blue", "red", "green")).to eq("red")
expect(ab_test("link_color", { "blue" => 0.01 }, "red" => 0.2)).to eq("red")
expect(ab_test("link_color", { "blue" => 0.8 }, { "red" => 20 })).to eq("red")
expect(ab_test("link_color", "blue", "red") do |alternative|
"shared/#{alternative}"
end).to eq("shared/red")
end
end
end
context "and preloaded config given" do
before do
Split.configuration.experiments[:link_color] = {
alternatives: [ "blue", "red" ],
}
end
it "uses first alternative" do
expect(ab_test(:link_color)).to eq("blue")
end
end
end
describe "finished" do
it "should not raise an exception" do
expect { ab_finished("link_color") }.not_to raise_error
end
it "should call db_failover_on_db_error proc with error as parameter" do
Split.configure do |config|
config.db_failover_on_db_error = proc do |error|
expect(error).to be_a(Errno::ECONNREFUSED)
end
end
expect(Split.configuration.db_failover_on_db_error).to receive(:call).and_call_original
ab_finished("link_color")
end
end
end
end
context "with preloaded config" do
before { Split.configuration.experiments = {} }
it "pulls options from config file" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
ab_test :my_experiment
expect(Split::Experiment.new(:my_experiment).alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(Split::Experiment.new(:my_experiment).goals).to eq([ "goal1", "goal2" ])
end
it "can be called multiple times" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: ["goal1", "goal2"]
}
5.times { ab_test :my_experiment }
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "other_opt" ])
expect(experiment.goals).to eq([ "goal1", "goal2" ])
expect(experiment.participant_count).to eq(1)
end
it "accepts multiple goals" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ],
goals: [ "goal1", "goal2", "goal3" ]
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([ "goal1", "goal2", "goal3" ])
end
it "allow specifying goals to be optional" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "other_opt" ]
}
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.goals).to eq([])
end
it "accepts multiple alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [ "control_opt", "second_opt", "third_opt" ],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.map(&:name)).to eq([ "control_opt", "second_opt", "third_opt" ])
end
it "accepts probability on alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 67 },
{ name: "second_opt", percent: 10 },
{ name: "third_opt", percent: 23 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
expect(experiment.alternatives.collect { |a| [a.name, a.weight] }).to eq([["control_opt", 0.67], ["second_opt", 0.1], ["third_opt", 0.23]])
end
it "accepts probability on some alternatives" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt", percent: 34 },
"second_opt",
{ name: "third_opt", percent: 23 },
"fourth_opt",
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.34], ["second_opt", 0.215], ["third_opt", 0.23], ["fourth_opt", 0.215]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "allows name param without probability" do
Split.configuration.experiments[:my_experiment] = {
alternatives: [
{ name: "control_opt" },
"second_opt",
{ name: "third_opt", percent: 64 },
],
}
ab_test :my_experiment
experiment = Split::Experiment.new(:my_experiment)
names_and_weights = experiment.alternatives.collect { |a| [a.name, a.weight] }
expect(names_and_weights).to eq([["control_opt", 0.18], ["second_opt", 0.18], ["third_opt", 0.64]])
expect(names_and_weights.inject(0) { |sum, nw| sum + nw[1] }).to eq(1.0)
end
it "fails gracefully if config is missing experiment" do
Split.configuration.experiments = { other_experiment: { foo: "Bar" } }
expect { ab_test :my_experiment }.to raise_error(Split::ExperimentNotFound)
end
it "fails gracefully if config is missing" do
expect { Split.configuration.experiments = nil }.to raise_error(Split::InvalidExperimentsFormatError)
end
it "fails gracefully if config is missing alternatives" do
Split.configuration.experiments[:my_experiment] = { foo: "Bar" }
expect { ab_test :my_experiment }.to raise_error(NoMethodError)
end
end
it "should handle multiple experiments correctly" do
experiment2 = Split::ExperimentCatalog.find_or_create("link_color2", "blue", "red")
ab_test("link_color", "blue", "red")
ab_test("link_color2", "blue", "red")
ab_finished("link_color2")
experiment2.alternatives.each do |alt|
expect(alt.unfinished_count).to eq(0)
end
end
context "with goals" do
before do
@experiment = { "link_color" => ["purchase", "refund"] }
@alternatives = ["blue", "red"]
@experiment_name, @goals = normalize_metric(@experiment)
@goal1 = @goals[0]
@goal2 = @goals[1]
end
it "should normalize experiment" do
expect(@experiment_name).to eq("link_color")
expect(@goals).to eq(["purchase", "refund"])
end
describe "ab_test" do
it "should allow experiment goals interface as a single hash" do
ab_test(@experiment, *@alternatives)
experiment = Split::ExperimentCatalog.find("link_color")
expect(experiment.goals).to eq(["purchase", "refund"])
end
end
describe "ab_finished" do
before do
@alternative_name = ab_test(@experiment, *@alternatives)
end
it "should increment the counter for the specified-goal completed alternative" do
expect { ab_finished({ "link_color" => ["purchase"] }) }
.to change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal2) }.by(0)
.and change { Split::Alternative.new(@alternative_name, @experiment_name).completed_count(@goal1) }.by(1)
end
end
end
end
<MSG> Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning
fix active experiments when experiment is on a later version
<DFF> @@ -449,6 +449,14 @@ describe Split::Helper do
expect(active_experiments.first[0]).to eq "def"
expect(active_experiments.first[1]).to eq alternative
end
+
+ it 'should show an active test when an experiment is on a later version' do
+ experiment.reset
+ expect(experiment.version).to eq(1)
+ ab_test('link_color', 'blue', 'red')
+ expect(active_experiments.count).to eq 1
+ expect(active_experiments.first[0]).to eq "link_color"
+ end
it 'should show multiple tests' do
Split.configure do |config|
| 8 | Merge pull request #331 from ndrisso/bug/show-active-experiments-with-versioning | 0 | .rb | rb | mit | splitrb/split |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.