[Haskell-cafe] ghc7 bug? associated type + implicit parameters

[Antoine Latter]

Someone on the ghc-users list might have an idea. I recommend filing a
bug report:

http://hackage.haskell.org/trac/ghc/wiki/ReportABug

Even if the behavior isn't wrong, in could use a better error message!

Antoine

2011/4/14 Michal Konečný <mikkonecny at googlemail.com>:
> Hi,
>
> I have stumbled across some strange behaviour in ghc7.
>
> The following compiles fine with ghc 6.12.3:
>
> {-# LANGUAGE ImplicitParams #-}
> {-# LANGUAGE TypeFamilies #-}
> {-# LANGUAGE NoMonomorphismRestriction #-}
>
> module Test where
>
> class C t where
>    type TF t
>    ttt :: TF t -> t
>
> b :: (C t, ?x :: TF t) => t
> b = ttt ?x
>
> but ghc7 says:
>
>    Could not deduce (?x::TF t)
>      arising from a use of implicit parameter `?x'
>    from the context (C t, ?x::TF t)
>      bound by the type signature for b :: (C t, ?x::TF t) => t
>      at Test.hs:13:1-10
>    In the first argument of `ttt', namely `?x'
>    In the expression: ttt ?x
>    In an equation for `b': b = ttt ?x
>
> Moreover, when I comment out the type declaration for b, it compiles and the
> inferred type for b is identical to the one in the above program:
>
> *Test> :t b
> b :: (C t, ?x::TF t) => t
>
> It feels to me like a bug but I am not entirely confident.  Any ideas?
>
> Michal
> --
> |o| Michal Konecny <mikkonecny at gmail.com>
> |o|    http://www-users.aston.ac.uk/~konecnym/
> |o|    office: (+42) (0)121 204 3462
> |o| PGP key http://www-users.aston.ac.uk/~konecnym/ki.aston
>
> --
> |o| Michal Konecny <mikkonecny at gmail.com>
> |o|    http://www-users.aston.ac.uk/~konecnym/
> |o|    office: (+42) (0)121 204 3462
> |o| PGP key http://www-users.aston.ac.uk/~konecnym/ki.aston
>
> --
> |o| Michal Konecny <mikkonecny at gmail.com>
> |o|    http://www-users.aston.ac.uk/~konecnym/
> |o|    office: (+42) (0)121 204 3462
> |o| PGP key http://www-users.aston.ac.uk/~konecnym/ki.aston
>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>