workspace
stringclasses 4
values | channel
stringclasses 4
values | text
stringlengths 1
3.93k
| ts
stringlengths 26
26
| user
stringlengths 2
11
|
|---|---|---|---|---|
elmlang | general | I think I get it, i’ll try that now. Thank you <@Carman> and <@Freda> | 2019-01-22T10:22:53.785000 | Allyn |
elmlang | general | :slightly_smiling_face: | 2019-01-22T10:22:55.785200 | Allyn |
elmlang | general | Yes, you are right | 2019-01-22T10:42:07.785300 | Lynne |
elmlang | general | A positive thing here is that you don't have to deal with `Maybe Profile` in your other pages | 2019-01-22T10:42:40.785500 | Lynne |
elmlang | general | So, I’ve been following the ‘life of a file’ approach (aka <https://www.reddit.com/r/elm/comments/5jd2xn/how_to_structure_elm_with_multiple_models/dbuu0m4/>), and I’ve ended up in a slightly awkward situation:
I split my model up with a union type (i.e. `model = FooModel FooRecord | BarModel BarRecord`),
then later I split up my `Msg` type too (i.e. `type Msg = FooMsg Foos | BarMsg Bars`).
That all seemed well and fine (using `Cmd.map` and `Html.map`), but:
I’m increasingly running in to situations where I have to add an `otherwise -> (model, Cmd.none)` to handle cases where the `Msg` no longer makes sense with the `Model`. Is there a way around this, is it a smell? | 2019-01-22T13:16:56.791500 | Bernardo |
elmlang | general | Yeah I'm curious about this too. Although I'm guessing there might not be a perfect solution.
elm-spa-example uses cases like:
```
_ ->
( model, Log.error )
```
(Log.error sends a message to a error tracking service) | 2019-01-22T13:24:09.793500 | Nana |
elmlang | general | <@Nana> funny you should reference elm-spa-example, I just found elm-kitchen, which says:
> This is a modest attempt at providing a simplistic yet opinionated Elm SPA application skeleton based on rtfeldman’s Elm Example SPA,
And has a similar smoking gun, here: <https://github.com/allo-media/elm-kitchen/blob/master/skeleton/src/Main.elm#L129> | 2019-01-22T13:27:22.794900 | Bernardo |
elmlang | general | This is a hard problem. You want some messages to only come in when your're in certain states. | 2019-01-22T13:28:38.796000 | Carman |
elmlang | general | The fact that messages are asynchronous makes it even trickier | 2019-01-22T13:28:54.796400 | Carman |
elmlang | general | Consider firing off a slow HTTP request when your model is in the `FooModel ...` state. While the request is happening in the background, the user clicks a button and transitions the app into the `BarModel ...` state. _After_ that has happened, the HTTP request completes, giving you a `FooMsg ...`. How do you handle the situation? | 2019-01-22T13:30:27.799900 | Carman |
elmlang | general | Heh, I think your use of ‘a hard problem’ is very fitting, if you squint at “state” and “transitions” it does start to look a lot like “cache invalidation”, and of course: <https://martinfowler.com/bliki/TwoHardThings.html> | 2019-01-22T13:32:36.803600 | Bernardo |
elmlang | general | Here's another example for food of thought. Imagine that you have a messaging app. No matter which conversation the user is viewing, the app can get different Msgs, including new incoming texts from other conversations. You probably do want the texts to be added to the convos, even if you are viewing another one at the moment. | 2019-01-22T13:33:18.804900 | Bert |
elmlang | general | Okay, it sounds like I’m not going crazy, and this is a ‘thing’. Has the Elm community come up with a name for it I wonder? :thinking_face: | 2019-01-22T13:34:00.805600 | Bernardo |
elmlang | general | I've heard "make invalid state transitions impossible" or variations on that | 2019-01-22T13:34:55.806500 | Carman |
elmlang | general | As far as I can tell, there is no way to do this at the type level | 2019-01-22T13:35:40.808000 | Carman |
elmlang | general | Yeah. It’s exactly the word ‘impossible’ which made me think I was doing something wrong. I’d gotten all giddy after watching <@Leonie>’s “making impossible states impossible” talk, and then was confronted with this ‘situation’ :grimacing: | 2019-01-22T13:36:04.808500 | Bernardo |
elmlang | general | <https://discourse.elm-lang.org/t/can-phantom-types-be-used-to-restrict-the-possible-transitions-in-a-state-machine/508> | 2019-01-22T13:36:15.808800 | Carman |
elmlang | general | <@Bert> yeah seems like the best is usually to have your "data stores" on your main model, rather than on a "page model"
which mirrors how you do things in React/Redux/Vue | 2019-01-22T13:37:50.811000 | Nana |
elmlang | general | How is redux different, sorry? | 2019-01-22T13:40:35.811300 | Danika |
elmlang | general | <@Danika> I meant, storing server data on a "page model" in Elm is like storing things on a component in React, whereas storing it on the main model is like storing it in Redux | 2019-01-22T13:42:57.812800 | Nana |
elmlang | general | Beware of creating invalid states and lots of maybes if you put all your data on the top-level model | 2019-01-22T13:46:06.813800 | Carman |
elmlang | general | <@Carman> hmm how would that create invalid states / lots of maybes? :thinking_face: | 2019-01-22T13:50:00.815600 | Nana |
elmlang | general | Imagine a multi-step form that asks for address info on page 1, payment info on page 2, and a coupon code on page 3. If you put it all on the top level you end up with something like:
```
type alias Model =
{ state : State, page : Page }
type Page
= AddressPage
| PaymentPage
| CouponPage
| SuccessPage
type alias State =
{ street : Maybe String
, city : Maybe String
, country : Maybe String
, cardNumber : Maybe Int
, couponCode : Maybe String
}
``` | 2019-01-22T13:58:16.821900 | Carman |
elmlang | general | This allows lots of invalid states (you shouldn't be able to go to payment if you don't have an address) and has lots of maybes.
Contrast with something like:
```
type alias Model =
{ page : Page, ... }
type alias Address = { street : String, city : String, country : String }
type Page
= AddressPage { street : Maybe String, city : Maybe String, country : Maybe String }
| PaymentPage { address : Address, cardNumber : Maybe Int }
| CouponPage { address : Address, cardNumber : Int, couponCode : Maybe String }
| SuccessPage { address : Address }
``` | 2019-01-22T13:58:21.822100 | Carman |
elmlang | general | <@Carman> ah, I was talking about server data, form data definitely makes more sense to store on the "page" | 2019-01-22T14:02:54.823800 | Nana |
elmlang | general | I think the same idea goes for server data. If there are some states where it shouldn't exist then it probably shouldn't be on the top-level model | 2019-01-22T14:04:19.824500 | Carman |
elmlang | general | ```
type alias Model = { profile : Maybe ProfileFromServer, page : Page }
type Page = SignUp | ProfilePage | ...
```
versus
```
type Page
= SignUp
| ProfilePage ProfileFromServer
| ...
``` | 2019-01-22T14:06:00.826100 | Carman |
elmlang | general | Aye, I guess the choice is between:
1. Use tagged unions (`|`) in your model, or:
2. Don’t. | 2019-01-22T14:09:31.827500 | Bernardo |
elmlang | general | `type Choice = TaggedUnion | Dont` | 2019-01-22T14:10:06.829000 | Rosalee |
elmlang | general | hmm, but typically you first navigate to a profile page and *then* load the data | 2019-01-22T14:10:07.829200 | Nana |
elmlang | general | 1. Necessitates that you might end up with a message which only makes sense on a ‘different’ side of the union (aka the problem I first came here asking about).
2. Eliminates 1., but introduces ‘invalid states’. | 2019-01-22T14:10:59.830900 | Bernardo |
elmlang | general | 2 does not eliminate 1, it just obscures it | 2019-01-22T14:11:53.831900 | Carman |
elmlang | general | Uh, whoops, I put the numbers the wrong way around :joy: | 2019-01-22T14:12:07.832200 | Bernardo |
elmlang | general | and how would `ProfilePage ProfileFromServer` work with navigating through urls? | 2019-01-22T14:12:27.832700 | Nana |
elmlang | general | The problem <@Bernardo> was asking about earlier (handling messages that are unexpected given the current state of the model) exists no matter how you structure the app | 2019-01-22T14:13:39.833500 | Carman |
elmlang | general | <@Carman> yeah. I think it’s interesting how it manifests itself in different ways. | 2019-01-22T14:14:50.834600 | Bernardo |
elmlang | general | agreed :slightly_smiling_face: | 2019-01-22T14:15:06.835000 | Carman |
elmlang | general | So yeah, it’s:
1. Use tagged unions (`|`), and know you’re forced to end up (probably) with something like `otherwise -> ( model, Cmd.none )`, or:
2. Don’t use tagged unions, so you can always ‘find’ the value for which a message concerns, but accept that your model will inevitably allow inconsistent states to exist. | 2019-01-22T14:17:00.837300 | Bernardo |
elmlang | general | I think definitely use tagged unions for pages and such though | 2019-01-22T14:17:39.838100 | Nana |
elmlang | general | Thanks for the chat everyone, that was really productive! I have to lunch now though. | 2019-01-22T14:18:32.839000 | Bernardo |
elmlang | general | In my experience, you generally end up having to do `otherwise -> ( model, Cmd.none )` regardless of which approach you take | 2019-01-22T14:24:55.842000 | Carman |
elmlang | general | With tagged unions (`|`) it's usually at the end of a flat case statement on all the tags. Without them it's usually a compound case on a variety of maybes and booleans. | 2019-01-22T14:25:48.842900 | Carman |
elmlang | general | btw, the way I'm dealing with my server data, I don't consider it "wrong" to have data which is not relevant to the current page
for example, if you're viewing the profile of User A, having data stored for User B just means that if you then go back to User B's page, you don't have to fetch the data again | 2019-01-22T14:29:42.844600 | Nana |
elmlang | general | and it's always stored by ID, so User A's data would never overwrite User B's data | 2019-01-22T14:30:55.845500 | Nana |
elmlang | general | that sort of data makes sense to store on the top-level because you always want it around regardless of what page you're on | 2019-01-22T14:32:11.846200 | Carman |
elmlang | general | app data != page data ^^ | 2019-01-22T14:33:00.846500 | Danika |
elmlang | general | Afternoon. Having difficulty integrating an elm Browser Element into an existing react app, mainly when it comes to configuring webpack. Is using react-elm-component package and elm-webpackloader the advisable way to go about all this. | 2019-01-22T15:18:38.849300 | Leah |
elmlang | general | Which Github issue were you looking at? Are there any discussions on the thread? I’ve noticed the same thing (filed a possibly different bug) and would like to follow it as much as possible. | 2019-01-22T17:44:54.849700 | Melodee |
elmlang | general | To be sure, there’s no way to pattern match against record syntax, is there? | 2019-01-22T17:47:09.850500 | Bernardo |
elmlang | general | I.e. I went from
```
Customers String (Maybe CustomerInfo)
``` | 2019-01-22T17:47:27.850600 | Bernardo |
elmlang | general | to
```
Customers { filter: String, modal: Maybe CustomerInfo }
``` | 2019-01-22T17:47:40.850800 | Bernardo |
elmlang | general | But now instead of:
```
Customers _ (Just _) -> False
``` | 2019-01-22T17:48:12.851000 | Bernardo |
elmlang | general | I have to:
```
Customers page -> case page.modal of
Just _ -> False
Nothing -> True
``` | 2019-01-22T17:48:42.851300 | Bernardo |
elmlang | general | Hmm, does Elm have `isJust`? :thinking_face: :haskell: | 2019-01-22T17:49:22.851500 | Bernardo |
elmlang | general | you can match on a record, but can't match on the values in the record directly, so
```
case myPoint of
{ x, y } -> ...
``` | 2019-01-22T17:49:24.851700 | Virgie |
elmlang | general | no, though `Maybe.Extra` probably does | 2019-01-22T17:49:46.851900 | Virgie |
elmlang | general | Oh, that’s good to know! So that would bind `x` to whatever is in `x` in the record? | 2019-01-22T17:52:24.852100 | Bernardo |
elmlang | general | It does, and `Maybe.Extra` does have it :tada: | 2019-01-22T17:53:55.852300 | Bernardo |
elmlang | general | ```
Customers { modal } -> isJust modal
``` | 2019-01-22T17:53:59.852500 | Bernardo |
elmlang | general | That’s a little nicer :sunglasses: | 2019-01-22T17:54:06.852700 | Bernardo |
elmlang | general | Thanks! :cookie: | 2019-01-22T17:54:18.852900 | Bernardo |
elmlang | general | it's a little more limited than haskell can do, but this probably makes the totallity checker much easier. Also cramming more stuff into patterns might be bad style anyway | 2019-01-22T17:55:17.853100 | Virgie |
elmlang | general | How do you upgrade an elm dependency | 2019-01-22T19:37:56.853800 | Dexter |
elmlang | general | The way I do it is try to remove the dependency manually from elm.json, then add it back again with elm install | 2019-01-22T19:48:39.854400 | Lindsey |
elmlang | general | as far as I know there's no elm command to remove or upgrade a dependency though. | 2019-01-22T19:50:39.856100 | Lindsey |
elmlang | general | Thanks, that sound a bit backwards, but I’ll give it a shot :slightly_smiling_face: | 2019-01-22T19:58:06.856600 | Dexter |
elmlang | general | I’ve been playing with the idea of writing a Dhall parser for Elm. My main reason is that I want to learn more about parsers but it would also be nice to use Dhall instead of JSON when communicating with a Haskell server. | 2019-01-22T20:23:34.858400 | Dione |
elmlang | general | I looked at this issue <https://github.com/elm/compiler/issues/1828> | 2019-01-23T01:58:40.858600 | Jae |
elmlang | general | Anyone know where we can see the changes for `0.19.0-bugfix6`? | 2019-01-23T03:39:17.859600 | Garnett |
elmlang | general | <#C0K384K4Y|news-and-links>, rtfeldman's post a little while ago | 2019-01-23T03:51:00.860200 | Bert |
elmlang | general | Thanks! | 2019-01-23T03:51:08.860400 | Garnett |
elmlang | general | Not in much detail though | 2019-01-23T03:51:21.860600 | Bert |
elmlang | general | The base repo is updated for 0.19 now. :slightly_smiling_face: | 2019-01-23T09:22:12.862700 | Jeanene |
elmlang | general | Yeah, that’s similar to what I saw :sleepy: | 2019-01-23T10:25:18.862900 | Melodee |
elmlang | general | thanks! | 2019-01-23T10:25:19.863100 | Melodee |
elmlang | general | any parcel or travis expert that gets whats going on here? <https://travis-ci.org/Razzeee/htbah-character-creator> | 2019-01-23T19:04:16.866100 | Desire |
elmlang | general | It seems you have to install `request` manually and add it to your `package.json`. | 2019-01-24T03:15:11.866900 | Timika |
elmlang | general | npm install request --save | 2019-01-24T03:15:17.867100 | Timika |
elmlang | general | but why? | 2019-01-24T04:24:06.867400 | Desire |
elmlang | general | One of your dependencies has a dependency that is no longer automatically installed, so you have to do that yourself. I don’t think it’s related to travis or parcel at all. I’ve just yesterday created a new parcel project from scratch and did not have to install that dependency. | 2019-01-24T04:26:23.867600 | Timika |
elmlang | general | You have a lot of webpack related stuff in your `devDependencies`, maybe it’s one of those? | 2019-01-24T04:26:46.867800 | Timika |
elmlang | general | <https://docs.npmjs.com/cli/ls.html> | 2019-01-24T04:27:03.868000 | Timika |
elmlang | general | Might give you a hint which package is causing this. | 2019-01-24T04:27:17.868200 | Timika |
elmlang | general | already tried, did not work but I'm guessing elm at the moment, that's why I'm confused | 2019-01-24T04:28:12.868400 | Desire |
elmlang | general | did try `npm ls request-promise` earlier | 2019-01-24T04:29:52.868600 | Desire |
elmlang | general | None | 2019-01-24T04:30:02.868800 | Desire |
elmlang | general | should have tried this | 2019-01-24T04:30:10.869200 | Desire |
elmlang | general | i think that was added in the new 0.19 rev6 | 2019-01-24T04:30:43.869400 | Desire |
elmlang | general | That could be the case. But I wonder why I did not get the issue yesterday. :thinking_face: | 2019-01-24T04:31:10.869600 | Timika |
elmlang | general | well I don't like the design :confused: | 2019-01-24T04:32:24.869800 | Desire |
elmlang | general | BTW, I DM D&D 5e, nice to see another pen and paper player here! :wave: | 2019-01-24T04:32:36.870000 | Timika |
elmlang | general | nice, started a d&d round last year | 2019-01-24T04:34:30.870200 | Desire |
elmlang | general | also some c'thulu | 2019-01-24T04:34:44.870400 | Desire |
elmlang | general | ```takodana:oslo-elm-day-19-cpu-emulation-demo malax$ npm ls request
[email protected] /Users/malax/projects/malax/oslo-elm-day-19-cpu-emulation-demo
└─┬ [email protected]
└─┬ [email protected]
└── [email protected]``` | 2019-01-24T04:35:14.870600 | Timika |
elmlang | general | that’s weird. It installed request automatically for me. | 2019-01-24T04:35:30.870800 | Timika |
elmlang | general | It’s a different binwrap version too. | 2019-01-24T04:35:56.871000 | Timika |
elmlang | general | maybe it's not the bugfix6 for me | 2019-01-24T04:36:36.871200 | Desire |
elmlang | general | Yeah, I just saw you don’t have that. | 2019-01-24T04:37:00.871400 | Timika |
elmlang | general | do you need `elm-webpack-loader` still when you use parcel? | 2019-01-24T04:37:21.871600 | Timika |
elmlang | general | I tend to install elm explicitly via npm. | 2019-01-24T04:37:33.871800 | Timika |
Subsets and Splits