Simon Peyton Jones <[email protected]> writes:

> |  I don't think this would work in the case where there are no fields
> |  initialised?
>
> Oh yes, silly me. I was thinking that then we wouldn’t need to look at
> 'labels' at all, but that's not true.
>
> Well, at least then I'd replace [PostTc id [FieldLabel] with (PostTc
> ConLike). This makes it like ConPatOut in HsPat. Then the (Located id)
> field is redundant (we can get it from the ConLike), but that’s only
> true after typechecking, so maybe simpler to keep both.
>
> It amounts to moving the call to conLikeFieldLabels from tcExpr
> (RecordCon ...) to dsExpr (RecordCon ...). A small thing but I think
> it'd be better.
>
Mathew, did you ever get to this?

Cheers,

- Ben

Attachment: signature.asc
Description: PGP signature

_______________________________________________
ghc-devs mailing list
[email protected]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

Reply via email to