workspace
stringclasses 4
values | channel
stringclasses 4
values | text
stringlengths 1
3.93k
| ts
stringlengths 26
26
| user
stringlengths 2
11
|
|---|---|---|---|---|
elmlang | general | and that `Resolver` parameter which the documentation does not make it very clear what it should be used for | 2019-04-19T12:14:38.482200 | Vilma |
elmlang | general | it maps a response to a result. This means you have to handle all the failure cases of a `Response` up-front | 2019-04-19T12:15:57.482800 | Virgie |
elmlang | general | sometimes I just feel like I don’t get the Elm architecture :sweat: | 2019-04-19T12:26:48.483100 | Vilma |
elmlang | general | Indeed previously it was simpler to concatenate two tasks together: ` ... |> Http.toTask |> Task.andThen ...` I have yet to see if this can be mapped directly with the new API | 2019-04-19T12:31:54.484800 | Hoa |
elmlang | general | I’ve been trying for a bit now but I just can’t figure it out | 2019-04-19T12:32:13.485200 | Vilma |
elmlang | general | the old `Http.toTask` function seemed to be very user-friendly and easy to understand and use | 2019-04-19T12:32:35.485700 | Vilma |
elmlang | general | I just can’t see another way to achieve what I’m trying without something similar | 2019-04-19T12:32:57.486100 | Vilma |
elmlang | general | because I want to send a message which will set some flags (e.g. loading flag to true) before the HTTP request runs | 2019-04-19T12:33:27.486700 | Vilma |
elmlang | general | There was some discussion about this limitation, but I don’t remember which solutions came up. I believe there was a package that builds on the new HTTP package which makes easier to deal with the kind of issues. | 2019-04-19T12:36:08.488100 | Hoa |
elmlang | general | I’m going to take a look | 2019-04-19T12:36:21.488300 | Vilma |
elmlang | general | <@Vilma> Should be this: <https://package.elm-lang.org/packages/jinjor/elm-req/latest/> | 2019-04-19T12:39:09.488800 | Hoa |
elmlang | general | (Found out on Elm Discourse) | 2019-04-19T12:39:19.489100 | Hoa |
elmlang | general | Some context: <https://discourse.elm-lang.org/t/using-task-to-send-http-requests/2696> | 2019-04-19T12:39:53.489300 | Hoa |
elmlang | general | thanks <@Hoa> going to read into this | 2019-04-19T12:40:52.489600 | Vilma |
elmlang | general | but even then from what I’ve read it seems creating a custom Cmd Msg is not ideal because it will be run twice via de runtime? | 2019-04-19T12:50:20.490200 | Vilma |
elmlang | general | Lemme see. | 2019-04-19T12:50:33.490400 | Hoa |
elmlang | general | <https://medium.com/elm-shorts/how-to-turn-a-msg-into-a-cmd-msg-in-elm-5dd095175d84> | 2019-04-19T12:50:41.490700 | Vilma |
elmlang | general | > 1. Cmd Msg* is inefficient
> By creating your own Cmd Msg you are making 2 trips through the elm runtime. Which is unnecessary. The command you are sending out, will end up in some other branch of our update function. | 2019-04-19T12:50:57.490900 | Vilma |
elmlang | general | then it recommends either calling `update` recursively or passing the message directly in the `update` branch you are in | 2019-04-19T12:51:50.491500 | Vilma |
elmlang | general | I think I can do that but will involve me doing some changes to my app | 2019-04-19T12:52:10.492000 | Vilma |
elmlang | general | Ah, that one is a classic | 2019-04-19T12:53:13.492200 | Hoa |
elmlang | general | But wait, you have to use the Elm runtime to do a HTTP call. | 2019-04-19T12:53:37.492700 | Hoa |
elmlang | general | The basic idea is that, instead of doing two calls and manage two `Cmd`’s you combine two (or more) tasks and finally manage a single `Cmd`. | 2019-04-19T12:55:12.494200 | Hoa |
elmlang | general | I got some helper functions here: <https://package.elm-lang.org/packages/prikhi/http-tasks/1.0.0/Http-Tasks> | 2019-04-19T12:55:26.494400 | Earnest |
elmlang | general | <@Virgie> Thanks, works great! | 2019-04-19T13:00:48.494900 | Jerilyn |
elmlang | general | Anyone in here familiar or have any examples (besides rogeriochavez/spades boilerplate) of using elm-return to slim down an `update` function? | 2019-04-19T15:46:48.496200 | Buffy |
elmlang | general | Whoah this is interesting...
<https://ellie-app.com/5jykFvVbcDra1> | 2019-04-19T16:21:52.497500 | Buffy |
elmlang | general | I don’t see anything special, it looks like the default app. Am I missing something? | 2019-04-19T16:24:45.497600 | Allyn |
elmlang | general | Msg is a record instead of a sum. | 2019-04-19T16:25:15.497800 | Buffy |
elmlang | general | I realized it's a bad idea in the process tho because the view has to construct the whole record which is the opposite of what I'm trying to do...
That or you have to pass in `Msg` to the view function which felt backwards too.
I went back to a sum type. | 2019-04-19T16:26:16.498000 | Buffy |
elmlang | general | Anyone ever tried modeling `Msg -> Model` with a Result type? | 2019-04-19T18:25:30.000100 | Buffy |
elmlang | general | Yes. Some people have done that. | 2019-04-19T19:03:16.001600 | Ashton |
elmlang | general | Any examples? I'm trying to make "Impossible States" in my update instead of defaulting to a `_ -> ...` case. | 2019-04-19T19:04:09.002700 | Buffy |
elmlang | general | Personally I havent seen much value in the approach, but I know at least a few people seem quite satisfied doing it. | 2019-04-19T19:04:48.004100 | Ashton |
elmlang | general | When you case against a pair you get Msg * Model number of states, but with an Either it'd be Msg + Model states which should map better to what I'm trying to do.
Well I might need a regular Sum instead of an either.
You can see an example of something close to what I'm doing with <https://github.com/rogeriochaves/spades/blob/master/boilerplate/src/Update.elm#L21>
Only I want to get rid of the defaulting that's required to "ignore" all the other page states by modeling the whole orchestration as a Sum instead of a Product.
<https://github.com/rogeriochaves/spades/blob/master/boilerplate/src/Update.elm#L21> | 2019-04-19T19:07:28.007700 | Buffy |
elmlang | general | Yeah, <@Leoma> has a return package that treats updates as returing a result-y kind of thing, which may error. I recall someone at my local Elm meet up showing off a big application where every update returned a `Result Error (Model, Cmd Msg)` | 2019-04-19T19:08:18.009200 | Ashton |
elmlang | general | I guess Im a critic of this, so I would much rather convince you not to follow this way than help connect you with them. For example, at least on the theoretical argument of “making impossible states impossible”, I dont think `Msg`/`Model` combinations are impossible states. | 2019-04-19T19:09:35.011500 | Ashton |
elmlang | general | 0 At least in my book, a `Msg` isnt a state. Its not memory that persists over time. It doesnt stay in your application. Its just a fleeting report of something that happened that your application needs to respond to.
1 Its a lot more possible than you might think. `Msg` come from the outside world. You dont know when or where they are coming from. Weird combinations can happen. I know this, in part, because I and others have put analytics in our applications that report impossible states that we _thought_ were impossible. | 2019-04-19T19:10:55.013900 | Ashton |
elmlang | general | First of all, I totally understand where you're coming from (or at least I think we understand each other).
Therefor, I'm really modeling this as a Product of Sums, where "Outside" messages are a part of the top level product, but then I want to mitigate someone's ability to something like `onClick Login` if they're already logged in, yeh?
EG: I have a `type Model = Authentication {email : String, password: String} | Main {name: String, etc : }` and a `type Msg = Authentication CredsMsg | Main MainMsg` and I'm trying to have an update that looks something like
```
update : Msg -> Model -> (Model, Cmd Msg)
update msg mdl =
case ??? of
Authentication message, Authentication model ->
Main message, Main model ->
```
which I have a version of that using a Tuple but then the compiler yells at me about not covering the `Authentication _, Main _` and `Main _, Authentication _` cases which are what I'm trying to avoid because they don't make sense. | 2019-04-19T19:16:46.019100 | Buffy |
elmlang | general | I hope what I'm saying makes sense... happy to paste a larger snippet if it'd be useful? | 2019-04-19T19:17:03.019600 | Buffy |
elmlang | general | I think I hear you. | 2019-04-19T19:19:12.019900 | Ashton |
elmlang | general | I think its okay to just have a case that goes to nothing happening. So, just in pure practical terms, I think:
```
(MainMsg subMsg, Authneitcation authModel) ->
(model, Cmd.none)
```
is an okay to way to go. | 2019-04-19T19:20:57.021800 | Ashton |
elmlang | general | The Tuple is the problem algebraically speaking... yes I _could_ use a `_ -> ...` to just ignore the bad cases at runtime... and it'd work but then if me or someone comes along later the compiler won't yell at us if we try to add a case like `(Authentication _, Main _) -> -- This case should be impossible to even construct`. | 2019-04-19T19:23:38.024700 | Buffy |
elmlang | general | `(a, b)` means there's `a * b` number of cases so if a and b are both type `Bool` that's 4 cases. But if I use a Sum of Bools ala:
```
type MsgModelThingMaybe = Auth Bool | Main Bool
```
The you have `Auth + Main` which is only two constructors to case on. `|` is type speak for addition algebraically speaking. Where a Tuple or an anonymous Product is type speak for multiplication. | 2019-04-19T19:27:33.029700 | Buffy |
elmlang | general | Yeah also I like to avoid `_ ->`. And to avoid incompleteness, you could just write out some of the non-sense cases. I think for practical purposes that could be the best way to handle this. | 2019-04-19T19:29:16.030400 | Ashton |
elmlang | general | Im still reluctant to call these combinations “nonsense” or “impossible”. They could happen. Maybe extremely rarely. Maybe theres no reason to handle them. But its still there. Its a logical possibility, and a very remote and trivial real possibility. | 2019-04-19T19:30:33.031600 | Ashton |
elmlang | general | Man loopty loop... I think for now I'm gonna settle on `Debug.todo` in that case that "shouldn't" exist. Oh bother... :bear: :honey_pot: | 2019-04-19T20:16:54.034400 | Buffy |
elmlang | general | A key thing is that messages are asynchronous and can show up late. So “impossible” stuff can happen pretty easily. | 2019-04-19T20:33:55.035400 | Dede |
elmlang | general | I'm aware messages are asynchronous at least in principle, what I'm asking about arises from me trying to learn <@Willodean>/elm-return lib...
I just searched it on the packages site and <@Ashton> you have a similar package...
What I'm trying to do (maybe without the whole Lens thing) is this... <https://package.elm-lang.org/packages/toastal/return-optics/latest>
But that's 0.18 and hasn't been upgraded it seems. | 2019-04-19T20:41:06.037800 | Buffy |
elmlang | general | <@Myrtle> isn't using a `_ -> ...` in examples from what I can tell... | 2019-04-19T20:41:54.038600 | Buffy |
elmlang | general | What is the challenge? | 2019-04-19T20:42:35.040000 | Willodean |
elmlang | general | Trying to use you're lib to facilitate composition of updates without resorting to a `_ -> Debug.todo` case... | 2019-04-19T20:43:23.040900 | Buffy |
elmlang | general | I started down this path following the spades boilerplate but every child component has a default case that NoOp's and I want to make those cases incorrect by construction. | 2019-04-19T20:44:21.042100 | Buffy |
elmlang | general | Want to avoid -> <https://github.com/rogeriochaves/spades/blob/master/boilerplate/src/Cats/Update.elm#L25> | 2019-04-19T20:44:40.042400 | Buffy |
elmlang | general | If you scroll to the bottom of the README here apparently there's a way to do it?
<https://package.elm-lang.org/packages/toastal/return-optics/latest> | 2019-04-19T20:45:04.042800 | Buffy |
elmlang | general | But that package isn't updated to 0.19 and I'm also not super sure I should need Lenses to accomplish this? | 2019-04-19T20:45:28.043300 | Buffy |
elmlang | general | That make any sense? | 2019-04-19T20:45:35.043600 | Buffy |
elmlang | general | You can't | 2019-04-19T20:47:26.043900 | Willodean |
elmlang | general | You are going to have a Noop case | 2019-04-19T20:47:37.044300 | Willodean |
elmlang | general | Either you just don't have that case, or you do. | 2019-04-19T20:48:03.044900 | Willodean |
elmlang | general | If you do, you do. Period | 2019-04-19T20:48:14.045300 | Willodean |
elmlang | general | Also, don't use `_ ->` in pattern matches | 2019-04-19T20:49:01.046100 | Willodean |
elmlang | general | Enumerate or lose exhaustivity checks | 2019-04-19T20:49:20.046700 | Willodean |
elmlang | general | This is what I'm trying to avoid. | 2019-04-19T20:49:21.046800 | Buffy |
elmlang | general | I'm trying to enumerate. But a Tuple (Anonymous Product, etc...) is algebraic multiplication and it's causing more states than are valid for my use case but trying to tie the Msg and Models together in a Sum is not working like I'm expecting it too. | 2019-04-19T20:50:14.047900 | Buffy |
elmlang | general | Is this relevant? <https://package.elm-lang.org/packages/bChiquet/elm-accessors/latest/Accessors-Library> | 2019-04-19T20:51:15.048500 | Buffy |
elmlang | general | Ahh | 2019-04-19T20:57:57.048700 | Willodean |
elmlang | general | You have numerous invalid cases due to space routing | 2019-04-19T20:58:19.049400 | Willodean |
elmlang | general | You are fucked bro | 2019-04-19T20:58:28.049700 | Willodean |
elmlang | general | The answer to this problem is injective type families or associated data families | 2019-04-19T20:58:51.050500 | Willodean |
elmlang | general | You can do this in Haskell but you are fucked in Elm | 2019-04-19T20:59:12.051000 | Willodean |
elmlang | general | Not sure what you mean by "space routing" and I've read through Sandi McGuire's Type Level programming book but haven't written anything with HKD or Type Families yet so still not super sure how they'd solve this problem.
Turning the function algebra into a Sum seems to be what I'm trying to do but I don't even know what I'm saying too... Ugh... I'm such a tweener right now it's painful. | 2019-04-19T21:02:00.055000 | Buffy |
elmlang | general | Clearly you can turn a `a -> b` into a Product ala `(a, b)` but not sure how to then say I only want `(Int, Int) | (Bool, Bool)` and not also `(Int, Bool) | (Bool, Int)`... :confused: | 2019-04-19T21:04:32.056400 | Buffy |
elmlang | general | I tried parameterizing the type constructor but that ended up polluting a bunch of shit `type ActionModel a b c d = PageOne a b | PageTwo c d` and didn't seem to do what was in my head. | 2019-04-19T21:07:39.058900 | Buffy |
elmlang | general | Plus `PageOne a b` is just another way of saying `(a, b)` anyways so I knew it wasn't right. | 2019-04-19T21:08:05.059600 | Buffy |
elmlang | general | They solve this problem because you can parameterize the second type by the first | 2019-04-19T21:08:19.060200 | Willodean |
elmlang | general | Or having them share a type | 2019-04-19T21:08:26.060500 | Willodean |
elmlang | general | I meant SPA routing | 2019-04-19T21:08:37.060800 | Willodean |
elmlang | general | Consider we have 3 types | 2019-04-19T21:09:38.062100 | Willodean |
elmlang | general | A route type | 2019-04-19T21:09:42.062300 | Willodean |
elmlang | general | A model type | 2019-04-19T21:09:46.062500 | Willodean |
elmlang | general | And a msg type | 2019-04-19T21:09:55.062900 | Willodean |
elmlang | general | With yah so far :slightly_smiling_face: | 2019-04-19T21:10:03.063100 | Buffy |
elmlang | general | update :: Msg (route :: Route) -> Model (route :: Route) -> Model (route :: Route) | 2019-04-19T21:11:10.064900 | Willodean |
elmlang | general | Now we are guaranteed that everything lines up | 2019-04-19T21:11:23.065300 | Willodean |
elmlang | general | You can't do this in Elm | 2019-04-19T21:11:49.065800 | Willodean |
elmlang | general | See my brain is trying to write that like `update : Msg Route -> Model Route -> Model Route` but that's not a thing right? But why? | 2019-04-19T21:12:45.067000 | Buffy |
elmlang | general | Because we want Route promoted or it gives us nothing | 2019-04-19T21:13:29.067600 | Willodean |
elmlang | general | Phantom types don't help here? I mean I can think of why it should work in my head but I can't see why it can't / won't work in Elm. | 2019-04-19T21:15:20.068500 | Buffy |
elmlang | general | Right | 2019-04-19T21:17:16.069400 | Willodean |
elmlang | general | Because you want this as the concrete case | 2019-04-19T21:17:38.070100 | Willodean |
elmlang | general | `Model 'PageOne` | 2019-04-19T21:17:53.070600 | Willodean |
elmlang | general | Not | 2019-04-19T21:18:29.071100 | Willodean |
elmlang | general | `Model Route` | 2019-04-19T21:18:35.071400 | Willodean |
elmlang | general | To make matters worse you will need to existentialize and deexistentialize to convince the compiler you are right about this | 2019-04-19T21:19:29.073100 | Willodean |
elmlang | general | `forall` is an issue | 2019-04-19T21:19:51.073700 | Willodean |
elmlang | general | I think the intuition of it is starting to make sense in my brain now... in other words, in Elm we're sorta just stuck with the `a x b` Tuple and `Debug.todo` crashing out if a bad case ever gets hit, yeh?
So where do Lenses fit into this whole thing?
Like how's this happening?
<https://package.elm-lang.org/packages/toastal/return-optics/latest>
Or is it just irrelevant? | 2019-04-19T21:21:26.075200 | Buffy |
elmlang | general | irrelevant | 2019-04-19T21:22:31.075600 | Willodean |
elmlang | general | frankly this is a problem that is a legit use case for dependent types | 2019-04-19T21:23:06.076200 | Willodean |
elmlang | general | however I would suggest that `Debug.todo` is a shitty answer | 2019-04-19T21:23:31.076600 | Willodean |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.