[GHC] #14733: Won't use (forall xx. f xx) with -XQuantifiedConstraints

[GHC]

#14733: Won't use (forall xx. f xx) with -XQuantifiedConstraints
-------------------------------------+-------------------------------------
        Reporter:  Iceland_jack      |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.5
      Resolution:                    |             Keywords:
                                     |  QuantifiedContexts wipT2893
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 Currently this is by-design, but we could change the design.

 These forall-constraints currently behave just like a local version of a
 top-level instance declaration, and those are all for classes.  You can't
 say
 {{{
 instance ... => f Int
 }}}
 although it'd make sense to do so.  So similarly you currently can't do
 that with the local-foralld constraints.


 * If we retain the restriction to class constraints only, we should reject
 the type signature with a civilised error message.

 * How bad is the restriction?  You can always say
 {{{
 class f a => C f a

 proof :: (forall xx. C f xx) => D (f a)
 }}}
   Would that do?  Or what do your use-cases look like?

 It'd be a moderate pain to generalise the facility, mainly because
 `InstEnv` (in which we look up instances) has class constraints as a
 deeply-baked-in assumption.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14733#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler