Seems reasonable to me 👍

We could put the binder length check in Language.PureScript.Sugar.CaseDeclarations I guess, although that name is a little misleading as it's actually for desugaring top level binders into case declarations currently.

Example error:

  case 42, 10, 12 of
    42, 10 -> doIt
    _ , _  -> return false

Error found:
in module CaseTest
at ...CaseTest.purs line 12, column 3 - line 16, column 1

  Binder list lengths differ in case alternatives:

    42, 10
    _, _

  Expecting 3 binders.

I could also do an error for each alternative that is incorrect instead of one for the whole case. Maybe that would be more useful, so you don't have to hunt for them.

That'd be nice - I'm not exactly sure how you'd determine it though... I guess there's a few rules that might work: pick the most common arity first, if there is no common arity then default to the first pattern?

I'd think whatever is in the case head would be treated as the canonical binder length.

Looks nice, thanks! Sorry for the slow response.

As for syntax, I think a comma separated list is probably best. The only other thing I can think of is to parenthesize, but that would almost certainly be more confusing for Haskellers since it's valid tuple syntax.

🚢 Thanks!

I've got to fix the test.

Ok, should be good to go. I changed it so it emits an error for each alternative with the incorrect number of binders

Error 1 of 2:

  in module CaseTest
  at CaseTest.purs line 13, column 5 - line 13, column 12

    Binder list length differs in case alternative:

      42, 10

    Expecting 3 binders.



  See https://github.com/purescript/purescript/wiki/Error-Code-CaseBinderLengthDiffers for more information,
  or to contribute content related to this error.

Error 2 of 2:

  in module CaseTest
  at CaseTest.purs line 14, column 5 - line 14, column 12

    Binder list length differs in case alternative:

      _, _

    Expecting 3 binders.



  See https://github.com/purescript/purescript/wiki/Error-Code-CaseBinderLengthDiffers for more information,
  or to contribute content related to this error.

@garyb Can we get this one merged in for 0.7.7? I'm leaving merges up to you at the moment since I know you had some ordering in mind. If so, @natefaubion, could you please merge master?

Is there going to be a 0.7.7? I figured we were pretty much there with 0.8 now anyway.

My order-specific PRs are in now, so merging whatever is fine again, although it looks like something I did means this will need rebasing first.

I wanted to discuss that actually. If there is nothing actually breaking in 0.8, maybe it's fine to skip 0.7.7.

Yeah, so far everything is backwards compatible - or enough so that SlamData still compiles anyway, probably a pretty good test 😄

Rebased.

Thanks!

Somehow this needs rebasing again. I'm not sure why, but I'll be sure to merge it immediately this time. Sorry...

Done. I think it was a whitespace issue.

👍 😌