[GHC] #14734: QuantifiedConstraints conflated with impredicative polymorphism?

[GHC]

#14734: QuantifiedConstraints conflated with impredicative polymorphism?
-------------------------------------+-------------------------------------
        Reporter:  Iceland_jack      |                Owner:  (none)
            Type:  bug               |               Status:  closed
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.5
      Resolution:  invalid           |             Keywords:
                                     |  QuantifiedConstraints 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):

 The difficulty with impredicativity is that, during inference, a
 unification variable may be unified with a polytype. Thus
 {{{
 f :: forall a. [a] -> [a]

 xs :: [forall b. b->b]

 ...(f xs)..
 }}}
 Here f really has a type argument, and that type is polymorphic.  We
 really have
 {{{
 ...(f @(forall b b->b) xs)...
 }}}
 The hard bit is infeering the implicit, invisible type arguments.

 But here we are the level of types.  With explicit arguments there should
 be nothing wrong with allowing polytypes. (It's a different matter for the
 implicit kind arguments, of course.)

 So... may it'd be OK.  I'm just not sure what a principled story is.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14734#comment:5>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler