[GHC] #9123: Need for higher kinded roles

[GHC]

#9123: Need for higher kinded roles
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                Owner:  goldfire
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler (Type    |              Version:  7.8.2
  checker)                           |             Keywords:  Roles,
      Resolution:                    |  QuantifiedConstraints
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):

 I think Coercible is a special case because `Coercible p q` and `p ~R# q`
 really are inter-convertible.  Not so for `Eq a` and `Ord a`.   And there
 might be many classes that happen to have `Eq a` (perhaps distantly
 transitively) as a superclass. Are we really going to search for a
 solution via all of those?  Now we are into backtracking.

 It can go wrong with `~R#` too.  Suppose we said
 {{{
 class (a ~R# b) => C a b where
   op :: a -> b
 }}}
 Now if we want `t1 ~R# t2`, one route might be by seeking `C t1 t2`.
 This way lies madness.

 I could live with a special case for `Corecible`/`~R#` and `~`/`~N#`.  But
 the general case looks swampy.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9123#comment:44>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler