workspace
stringclasses 4
values | channel
stringclasses 4
values | text
stringlengths 1
3.93k
| ts
stringlengths 26
26
| user
stringlengths 2
11
|
|---|---|---|---|---|
elmlang | general | Can you define a different decoder that reads from the correct field and use that instead? | 2019-01-10T14:46:33.065800 | Carman |
elmlang | general | Unfortunately that won't work when the server-side GraphQL schema changes and we need to regenerate code | 2019-01-10T14:47:22.066400 | Shawanna |
elmlang | general | The required change is predictable, but the only way I can see to really handle it now is to get a string body back, do the substitution, and then continue | 2019-01-10T14:48:20.067300 | Shawanna |
elmlang | general | What about modifying the code-gen script to do a find-and-replace at the very end? | 2019-01-10T14:52:38.068300 | Carman |
elmlang | general | that would end up breaking it for other, correct, backends | 2019-01-10T14:53:00.068600 | Shawanna |
elmlang | general | I could do that and then just have two sets of generated code | 2019-01-10T14:53:40.069000 | Shawanna |
elmlang | general | hmmm maybe your find-and-replace could do something a bit fancier like turning `field "name" string` into `oneOf [field "name" string, field "email" string]` ? | 2019-01-10T14:57:39.072000 | Carman |
elmlang | general | ahh, yeah, that's a good idea | 2019-01-10T14:58:49.073100 | Shawanna |
elmlang | general | <@Carman> hey there! For reference, here is the issue <@Shawanna> is talking about: <https://github.com/dillonkearns/elm-graphql/issues/106>. It's a bit hard to describe concisely. But basically there is this strange clause in the GraphQL Spec that declares certain perfectly unambiguous queries to be invalid: <https://facebook.github.io/graphql/draft/#example-54e3d>
Some GraphQL team members were nice enough to describe the rationale, which is basically that they wanted to eliminate a case where, say, a Java GraphQL code generator would have a clash of two fields with the same name but different types. So they made them invalid.
All the relevant details are linked to in that issue. Like I said, pretty complicated so it's a bit hard to describe it concisely! | 2019-01-10T15:01:27.075600 | Sade |
elmlang | general | Hey <@Sade>. It turns out for my case, the only alias that really matters is the alias of the actual mutation operation, so I've got a pretty dirty way of stripping it, now I'm just looking to add it back in on completion, since I can guarantee there will never be a need for it. | 2019-01-10T15:03:12.078300 | Shawanna |
elmlang | general | Hence the idea of renaming a field in the decode pipeline | 2019-01-10T15:03:36.079100 | Shawanna |
elmlang | general | The unfortunate thing is that it seems like a couple of frameworks don't handle valid GraphQL queries correctly when they have valid (but unusual) field aliases like this:
```
{
search(query: "Elm") {
...on Repo {
aliasForStringId: id
}
...on Issue {
aliasForIntId: id
}
}
}
``` | 2019-01-10T15:05:04.079800 | Sade |
elmlang | general | ^ And by the way, this is exactly the case where you need a field alias to ensure that this query is valid (even though it would be unambiguous without the aliases). But since in practice field aliases are rarely used, some frameworks have bugs around them :confused: | 2019-01-10T15:05:56.080600 | Sade |
elmlang | general | Yeah, it's an unfortunate intersection of little used edge cases | 2019-01-10T15:06:16.081000 | Shawanna |
elmlang | general | AppSync handles the spec correctly (I think), it just fails to automagically trigger the subscription event unless a mutation matches exactly by name. Hence, I only need to strip that alias, then I can parse the result w/ field aliases no-problem | 2019-01-10T15:07:15.082400 | Shawanna |
elmlang | general | > It turns out for my case, the only alias that really matters is the alias of the actual mutation operation
<@Shawanna> I'm not sure I'm following, could you clarify with an example maybe? | 2019-01-10T15:07:27.082700 | Sade |
elmlang | general | Sure | 2019-01-10T15:07:33.082900 | Shawanna |
elmlang | general | Here's an example subscription
```
subscription {
showStartedAt {
showState3523593770: showState
comedians2741155849: comedians
time3832528868: time
date3832528868: date
}
}
``` | 2019-01-10T15:07:55.083300 | Shawanna |
elmlang | general | That gets wired up in AppSync so that it automatically triggers on a specific mutation | 2019-01-10T15:08:32.083900 | Shawanna |
elmlang | general | So when I send a mutation like this:
```
mutation {
startShow3832717217: startShow(date: "2019-02-02", time: "02:00 pm") {
showState3523593770: showState
comedians2741155849: comedians
time3832528868: time
date3832528868: date
}
}
``` | 2019-01-10T15:08:43.084200 | Shawanna |
elmlang | general | It should trigger it, but doesn't, because of the `startShow3832717217` alias | 2019-01-10T15:09:17.084700 | Shawanna |
elmlang | general | So, if I strip that alias, send the response, and then add it back in, everything lines up | 2019-01-10T15:09:43.085600 | Shawanna |
elmlang | general | I just realized I may be able to use the alias in the AppSync GraphQL annotations that specify the wiring, but I'll have to test it | 2019-01-10T15:10:26.087100 | Shawanna |
elmlang | general | Ah, okay, it's very helpful to see the specific example, thanks for that <@Shawanna>!
So this is actually a totally different use case then. This doesn't have to do with that corner-case of the GraphQL Spec that I was describing. This is actually something that `dillonkearns/elm-graphql` does in order to ensure that you don't have clashing fields if you query the same thing multiple times. | 2019-01-10T15:10:34.087500 | Sade |
elmlang | general | That'd certainly make things simpler | 2019-01-10T15:10:35.087600 | Shawanna |
elmlang | general | yes | 2019-01-10T15:10:47.087800 | Shawanna |
elmlang | general | The feature makes 100% sense | 2019-01-10T15:10:58.088000 | Shawanna |
elmlang | general | Yeah, totally, the aliasing is all valid according to the GraphQL spec. The only reason I bring it up is because this one is actually harder to find a way to remove. Because say you had multiple calls to `startShow` in your `dillonkearns/elm-graphql` query which each have different arguments... then you would need the hashed aliases that it uses (based on the arguments passed in). | 2019-01-10T15:12:34.089700 | Sade |
elmlang | general | So anyway, there's not really anything that can be done to remove that. As I describe in <https://medium.com/@dillonkearns/how-elm-guides-towards-simplicity-3d34685dc33c> | 2019-01-10T15:12:59.090400 | Sade |
elmlang | general | Agreed. I happen to know that, in my use-case, that won't ever be a concern, so I copied only the `Graphql.Http` part of `elm-graphql` into my project and rework `toReadyRequest` to strip the alias and capture it's name for later re-insertion | 2019-01-10T15:14:41.092200 | Shawanna |
elmlang | general | I'd been hoping to write something that could generally transform a `Decoder` to rename a field, but looking at the source for `Json.Decode` it doesn't seem likely | 2019-01-10T15:15:15.093000 | Shawanna |
elmlang | general | It's a fragile approach, unfortunately | 2019-01-10T15:15:50.093900 | Shawanna |
elmlang | general | Yeah, I wouldn't imagine that tweaking Json.Decode would be a fruitful path. But copying the code into `vendor` in your project wouldn't be too terrible. Obviously it's far from ideal, but it should do the trick. | 2019-01-10T15:16:14.094400 | Sade |
elmlang | general | In any case, I think I just found an AWS side work-around | 2019-01-10T15:16:32.095100 | Shawanna |
elmlang | general | The good news is that there's really just a single pinch point for determining the alias names. So all you need to do is tweak that any time you pull down updated source code. | 2019-01-10T15:16:38.095200 | Sade |
elmlang | general | Oh, good! | 2019-01-10T15:16:48.095400 | Sade |
elmlang | general | If your curious, basically in my mutation schema I just define a dummy operation:
```
type Mutation {
startShow(date: String!, time: String!): Show!
startShow3832717217(date: String!, time: String!): Show!
}
```
And wire the annotation to both of them, then it ends up routing things correctly | 2019-01-10T15:17:46.096400 | Shawanna |
elmlang | general | Since the usage pattern is simple that isn't too painful to keep track of | 2019-01-10T15:18:40.097300 | Shawanna |
elmlang | general | Thanks for the help <@Sade> and <@Carman> | 2019-01-10T15:18:56.097700 | Shawanna |
elmlang | general | Oh my! Well, whatever gets the job done, I suppose! :man-shrugging: Glad you found a solution. Would you mind posting that on the Github issue when you get a chance? | 2019-01-10T15:21:04.097800 | Sade |
elmlang | general | Ahh dang, nevermind. I'm dumb and didn't consider the hashing was on data, not just keys. Ok, back to the hacky plan then | 2019-01-10T15:31:29.098900 | Shawanna |
elmlang | general | Anyone have some pointers on how I might use `elm/parser` to parse a string like "45.4.5.6", where the first number must be between 1-64, and the numbers following the first (4.5.6) are optional but the values of which must be within the range of 1-6? | 2019-01-10T16:55:28.101500 | Santina |
elmlang | general | <@Santina> roughly
1. Decode an integer with `int`
2. use `Decode.andThen` to get access to the parsed value
3. validate
```
<http://Parser.int|Parser.int> |> Parser.andThen (\value ->
if value >= 1 && value <= 64 then
Parser.loop [] dotDigit
|> Parser.map (\digits -> (value, digits)
else
problem "out of range"
```
4. use `Parser.loop` to parse the optional final digits. This decoder will probably use `chompIf` a bunch.
5. put it all together | 2019-01-10T17:02:38.101900 | Virgie |
elmlang | general | thank you!!! this is very helpful | 2019-01-10T18:25:03.102100 | Santina |
elmlang | general | dotDigit is a function I'd write, correct? | 2019-01-10T18:26:00.102300 | Santina |
elmlang | general | Hi folks! Anyone using parcel bundler with elm? I was really happy, everything working perfectly.
Today I run into a problem and cannot see what I am doing wrong. I created a port and it seems after compiling it's not there.
When I compile with `elm make` directly the port shows up as expected | 2019-01-10T18:42:57.105700 | Pauletta |
elmlang | general | <@Pauletta> are you using that port within Elm? | 2019-01-10T18:48:23.106300 | Earlean |
elmlang | general | Dead code elimination will remove unused ports | 2019-01-10T18:48:52.107000 | Earlean |
elmlang | general | A prod ready web-app developed in 100% elm and elm-ui , try it out @ <https://mhctrucksmartsearch.azurewebsites.net/> | 2019-01-10T19:15:41.107200 | Ron |
elmlang | general | :wave: looking for a suggestion on how to decode something in json. let's say i have
```
type alias RowItem = Result String (Maybe String)
```
and my json looks like
```
{
"userID": "123",
"userIDError": ""
}
```
i want to make a `RowItem` decoder that depends on both of those keys.
*examples*
```
{ "userID": "", "userIDError": "Something bad happened" }
// -> Err "Something bad happened"
```
```
{ "userID": "123", "userIDError": "" }
// -> Ok (Just "123")
```
```
{ "userID": "", "userIDError": "" }
// -> Ok Nothing
```
does that make sense? i was playing around with `Decode.andThen`, but i couldn't figure out how get the behavior i want. | 2019-01-10T21:46:56.113400 | Dee |
elmlang | general | hm, maybe `Decode.oneOf` will do the trick | 2019-01-10T21:52:52.114000 | Dee |
elmlang | general | Decode the first field `andThen` decode the second field `andThen` use both values to make your final type | 2019-01-10T23:48:02.115000 | Earnest |
elmlang | general | <@Dee> | 2019-01-10T23:48:06.115200 | Earnest |
elmlang | general | Hello. I created a 3 days training to help members of my team to start with elm. One of them is concerned with the language maturity. Have you any useful info I could give about this subject? Do we know how far we are from a 1.0 version? (until then I extracted a few numbers of the state of elm 2018 poll, about the use of elm in production) | 2019-01-11T01:20:18.118300 | Allyn |
elmlang | general | there's no established milestone for what 1.0 means for elm | 2019-01-11T02:00:39.118600 | Lashawnda |
elmlang | general | when we need to argue the degree to which elm is production ready we usually do so in terms of the abundance of cases where elm is successfully being used in production | 2019-01-11T02:01:27.118800 | Lashawnda |
elmlang | general | That's the way 've taken so far | 2019-01-11T02:03:11.119000 | Allyn |
elmlang | general | another data point that i like is that there are now 4 independent conferences dedicated entirely to elm in 3 different countries | 2019-01-11T02:03:54.119200 | Lashawnda |
elmlang | general | 4 conferences if I'm not wrong | 2019-01-11T02:04:13.119400 | Allyn |
elmlang | general | Elm in the spring, Oslo, Elm Europe and Elm Conf | 2019-01-11T02:04:46.119600 | Allyn |
elmlang | general | Do you have any example of enterprises I could use, in addition to NoRedInk, FeatureSpace, IBM (I referred to the post they published on elm discourse)? | 2019-01-11T02:05:06.119800 | Allyn |
elmlang | general | 1.0 shouldn't really matter. React was 0.x for a few years while people were totally happy to use it in production. Then at some point they got tired of the 1.0 question and started numbering releases with x.0 instead... | 2019-01-11T02:06:25.120000 | Bert |
elmlang | general | I definitely agree with that, and it was not the point of my co-worker | 2019-01-11T02:07:28.120200 | Allyn |
elmlang | general | But I have a vague memoy of reading somewhere (maybe here) that elm 0.19 was near of what we could call a 1.0 release in terms of language design | 2019-01-11T02:10:32.120500 | Allyn |
elmlang | general | I was wondering if any concrete info could confirm this | 2019-01-11T02:11:19.120700 | Allyn |
elmlang | general | <@Colby> thanks for the suggestion of mentionning conferences, I'll do it | 2019-01-11T02:12:10.120900 | Allyn |
elmlang | general | regarding the version number, the official line is that we're not concerned right now with what a finished iteration that a 1.0 version might imply would look like, and we're just continuing to make it nicer for the foreseeable future | 2019-01-11T02:13:50.121200 | Lashawnda |
elmlang | general | I like this line :slightly_smiling_face: | 2019-01-11T02:14:27.121400 | Allyn |
elmlang | general | As for the companies, there's this list at least: <https://github.com/jah2488/elm-companies> | 2019-01-11T02:20:40.121600 | Bert |
elmlang | general | <@Earnest> that worked! thanks a lot!
```
decodeRowItem : String -> Decode.Decoder RowItem
decodeRowItem field =
Decode.field (field ++ "Error") Decode.string
|> Decode.andThen
(\err ->
Decode.field field Decode.string
|> Decode.andThen
(\val ->
Decode.succeed (mapRowItem err val)
)
)
```
```
> json = """{"userID":"", "userIDError":"Something bad happened."}"""
"{\"userID\":\"\", \"userIDError\":\"Something bad happened.\"}" : String
> Decode.decodeString (decodeRowItem "userID") json
Ok (Err ("Something bad happened.")) : Result Decode.Error (Result String (Maybe String))
> json = """{"userID":"123", "userIDError":""}"""
"{\"userID\":\"123\", \"userIDError\":\"\"}" : String
> Decode.decodeString (decodeRowItem "userID") json
Ok (Ok (Just "123")) : Result Decode.Error (Result String (Maybe String))
> json = """{"userID":"", "userIDError":""}"""
"{\"userID\":\"\", \"userIDError\":\"\"}" : String
> Decode.decodeString (decodeRowItem "userID") json
Ok (Ok Nothing) : Result Decode.Error (Result String (Maybe String))
``` | 2019-01-11T03:32:43.122900 | Dee |
elmlang | general | *Poll:* Do you use a single `Id` type for everything, or do you use `User.Id` `Post.Id` `Comment.Id` etc?
Vote :+1: for single `Id` or :-1: if you use different Ids, also like to hear what you think :slightly_smiling_face: | 2019-01-11T03:54:44.124900 | Nana |
elmlang | general | is this still the latest version available on npm? 0.19.0-bugfix2 | 2019-01-11T04:09:26.125300 | Selene |
elmlang | general | there is also this, which is just logos <https://docs.google.com/presentation/d/1LM_W2BRs_ItT-SPDe70C10cbwhGNHGQlJ1fVnAdnRIY/edit#slide=id.g400e8d3ac3_0_208> | 2019-01-11T04:18:18.125700 | Desire |
elmlang | general | Yes, I'm using it, the port is there when compiling with elm make. I think it might be the way parcel calls the compiler, I'll try to dig a bit more. Thank you! | 2019-01-11T04:27:17.126000 | Pauletta |
elmlang | general | for an elm18 project, that also needs elm dependencies from github instead of the officially hosted packages...
`elm-github-install` or `elm-groove` .. or? any recommendations? | 2019-01-11T05:20:49.127800 | Earnestine |
elmlang | general | I voted :-1:, but if your data comes from a traditional relational DB a single `Id` makes sense :slightly_smiling_face: | 2019-01-11T05:29:04.128000 | Hoa |
elmlang | general | While dealing with third-party API situation is less clear | 2019-01-11T05:30:32.128200 | Hoa |
elmlang | general | I use phantom types when I can
```
type Id a
= Id String
```
Not sure if thats a :thumbsup: or :thumbsdown: . Its one type `Id a`, but it comes in forms `Id User`, `Id Comment`, `Id Post` | 2019-01-11T06:52:15.128700 | Ashton |
elmlang | general | I had no idea such things existed | 2019-01-11T06:52:28.129100 | Danika |
elmlang | general | FYI: For 0.19 there’s <https://github.com/Skinney/elm-git-install> | 2019-01-11T06:57:42.129400 | Hoa |
elmlang | general | <@Ashton> that's pretty cool :open_mouth: so you get the same safety but without the boilerplate | 2019-01-11T06:59:50.129600 | Nana |
elmlang | general | :exploding_head: | 2019-01-11T07:01:18.129900 | Danika |
elmlang | general | point of it today ? can you still crate some “ugly magic” like native modules ? | 2019-01-11T07:02:12.130500 | Liza |
elmlang | general | No | 2019-01-11T07:02:23.130700 | Danika |
elmlang | general | for private repo only ? | 2019-01-11T07:03:07.131900 | Liza |
elmlang | general | Yes pretty much | 2019-01-11T07:04:20.133500 | Danika |
elmlang | general | rip native modules :c | 2019-01-11T07:04:25.133700 | Danika |
elmlang | general | interesting is there any hacky way how to crate native modules, and run locally - just for learning exp | 2019-01-11T07:05:10.134400 | Liza |
elmlang | general | I would say not in 0.19, no | 2019-01-11T07:05:24.134600 | Danika |
elmlang | general | Ports are the only option now | 2019-01-11T07:05:55.135100 | Danika |
elmlang | general | but how i could become contributor some day, if i have no way to hack / learn ? | 2019-01-11T07:06:37.135700 | Liza |
elmlang | general | A contributor to what? | 2019-01-11T07:07:33.136100 | Danika |
elmlang | general | `elm-community/*` `elm-explorations/*` `elm/*` | 2019-01-11T07:08:16.136700 | Liza |
elmlang | general | <@Ashton> could you give an example how to create such Id User for instance?
Is it something like the following?
```
userId: String -> Id User
userId s = Id s
``` | 2019-01-11T07:12:55.138600 | Elyse |
elmlang | general | > So in 0.19 only the elm-lang and elm-explorations organizations can compile and publish kernel code or effect managers. It *will not be available in applications* or other packages. I hope the history I wrote, and the posts here 303 and here 214 clarify the design trade-offs that went into this.
I believe the intention is for lang/explorations to eventually drop kernel code all together. Although presumably there is some mechanism built into the language to allow those developers to actually work on those packages… A private fork perhaps?
In any case the kernel code always lives in `src/Elm/Kernel/yourJsHere.js` so try it and see what the compiler shouts about I guess. | 2019-01-11T07:13:28.139100 | Danika |
elmlang | general | based on what i sow - your package also must have some magic organization stuff, an as i know `elm-lang` and `elm-explorations` is hardcoded into compielier | 2019-01-11T07:16:16.140300 | Liza |
elmlang | general | so - sounds like some “little” magic how to compile it, and then link it as dependency.. | 2019-01-11T07:17:05.141100 | Liza |
elmlang | general | and looking into how JS looks… it is not really pleasure work | 2019-01-11T07:17:49.141700 | Liza |
elmlang | general | Although I disagree with the basic premise, the point is you’re not supposed to be writing kernel code in your elm apps. It was never a supported feature for users, just an undocumented api designed just for core. If it works for lang/explorations then there _will_ be a way to get it to work locally, but it’s not a useful or productive “learning” process because there will eventually be no kernel code left to write. | 2019-01-11T07:25:00.143400 | Danika |
elmlang | general | It’s not a pleasure to work, because it’s a hack until WebAssembly support is complete | 2019-01-11T07:25:25.143900 | Danika |
Subsets and Splits