Int value implicitly converted to String?

[simonewebdesign]

I'm using evancz/elm-http to send an HTTP request, here's the code:

deleteArtist : Int -> Task Http.Error ()
deleteArtist id =
  Http.send Http.defaultSettings
    { verb = "DELETE"
    , headers = []
    , url = "http://localhost:4000/api/artists/" ++ id
    , body = Http.empty
    }
  `andThen` (\_ -> Signal.send actions.address (DeleteArtist id))
  `onError` (\error -> Signal.send actions.address NoOp)

I get a compile error:

The argument to function `DeleteArtist` is causing a mismatch.

130│                                                 DeleteArtist id)
                                                                  ^^
Function `DeleteArtist` is expecting the argument to be:

    Int

But it is:

    String

I found a workaround, which is to move the Http.Request details to its own function and pass the id to it:

deleteRequest : String -> Http.Request
deleteRequest id =
  { verb = "DELETE"
  , headers = []
  , url = "http://localhost:4000/api/artists/" ++ id
  , body = Http.empty
  }

deleteArtist : Int -> Task Http.Error ()
deleteArtist id =
  Http.send Http.defaultSettings (deleteRequest (toString id))
  `andThen` (\_ -> Signal.send actions.address (DeleteArtist id))
  `onError` (\error -> Signal.send actions.address NoOp)

I'm guessing that, in the first example, ++ is somehow changing the parameter type to String.
Is this expected behavior?
My expectation would be that the id parameter remains of type Int all the time.

Thanks in advance for any clarification.

[jvoigtlaender]

In a type correct program, id will have only one type. But your original program is not type correct. You use id as a String in one line, but as an Int in another line.

One could wish for the compiler to give you a clearer message about this. See elm/error-message-catalog#70 and related issues.

But the main point here is that no silent conversion happens (what you feared). There is no bug in compiler or libraries here, just an imperfect error message produced by the compiler.

[jvoigtlaender]

In a type correct program, id will have only one type. But your original program is not type correct. You use id as a String in one line, but as an Int in another line.

One could wish for the compiler to give you a clearer message about this. See elm/error-message-catalog#70 and related issues.

But the main point here is that no silent conversion happens (what you feared). There is no bug in compiler or libraries here, just an imperfect error message produced by the compiler.

[jvoigtlaender]

The way you should probably have written your function is:

deleteArtist : Int -> Task Http.Error ()
deleteArtist id =
  Http.send Http.defaultSettings
    { verb = "DELETE"
    , headers = []
    , url = "http://localhost:4000/api/artists/" ++ toString id
    , body = Http.empty
    }
  `andThen` (\_ -> Signal.send actions.address (DeleteArtist id))
  `onError` (\error -> Signal.send actions.address NoOp)

[jvoigtlaender]

The way you should probably have written your function is:

deleteArtist : Int -> Task Http.Error ()
deleteArtist id =
  Http.send Http.defaultSettings
    { verb = "DELETE"
    , headers = []
    , url = "http://localhost:4000/api/artists/" ++ toString id
    , body = Http.empty
    }
  `andThen` (\_ -> Signal.send actions.address (DeleteArtist id))
  `onError` (\error -> Signal.send actions.address NoOp)

[simonewebdesign]

Yes, that worked, thank you.
Looks like I got very confused by the error message, so it's probably worth improving it.

Alternatively I think would be nice to have an implicit toString call every time one tries to append some value to a String. But I don't know if that's feasible or even if makes sense at all.

[simonewebdesign]

Yes, that worked, thank you.
Looks like I got very confused by the error message, so it's probably worth improving it.

Alternatively I think would be nice to have an implicit toString call every time one tries to append some value to a String. But I don't know if that's feasible or even if makes sense at all.

[jvoigtlaender]

Looks like I got very confused by the error message, so it's probably worth improving it.

Yes, that's what the issue in the https://github.com/elm-lang/error-message-catalog/ repo I linked to is about.

So can this issue here be closed?

[jvoigtlaender]

Looks like I got very confused by the error message, so it's probably worth improving it.

Yes, that's what the issue in the https://github.com/elm-lang/error-message-catalog/ repo I linked to is about.

So can this issue here be closed?

[simonewebdesign]

Sure, thanks again for the help :)

[simonewebdesign]

Sure, thanks again for the help :)