[Haskell-cafe] Does GADTs imply ExistentialQuantification?

[Ömer Sinan Ağacan]

This actually looks like a bug in the implementation. If you look at the
"implied flags" list for GADTs it only implies GADTSyntax and MonoLocalBinds
[1]. I'm guessing that we're missing a check in the type checker code for
constructors when the syntax is existential syntax (I'm not sure if we can
distinguish a constructor with existential syntax vs. GADT syntax in type
checker though).

[1]: https://github.com/ghc/ghc/blob/c523525b0e434d848f6e47ea3f9a37485965fa79/compiler/main/DynFlags.hs#L4366-L4367

Ömer

David Feuer <david.feuer at gmail.com>, 27 Ağu 2018 Pzt, 04:56 tarihinde
şunu yazdı:
>
> I'm not sure. Once you have GADTs, ExistentialQuantification is just syntax. So from one angle, maybe it doesn't matter too much. On the other hand, that syntax, whole traditional, is not the clearest in the world. One could easily imagine a Haskell implementation that chooses only to support GADT syntax for existentials. So maybe it is a bug.
>
> On Aug 26, 2018 9:43 PM, "Cosmia Fu" <cosmiafu at gmail.com> wrote:
>
> Hey everyone,
>
> I found that it seems that GADTs implies ExistentialQuantification, but not mentioned in the manual. Is it a bug?
> https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#extension-GADTs
>
> GHCi, version 8.4.3: http://www.haskell.org/ghc/  :? for help
> Prelude> :set -XGADTs -XExplicitForAll
> Prelude> data Q = forall a. Q a
> Prelude>
>
> ----
>
> Cosmia Fu
> _______________________________________________
> Haskell-Cafe mailing list
> To (un)subscribe, modify options or view archives go to:
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> Only members subscribed via the mailman list are allowed to post.
>
>
> _______________________________________________
> Haskell-Cafe mailing list
> To (un)subscribe, modify options or view archives go to:
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> Only members subscribed via the mailman list are allowed to post.