[GHC] #11715: Constraint vs *

[GHC]

#11715: Constraint vs *
-------------------------------------+-------------------------------------
        Reporter:  bgamari           |                Owner:
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.2.1
       Component:  Compiler (Type    |              Version:  8.0.1-rc1
  checker)                           |
      Resolution:                    |             Keywords:  Typeable
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 int-index):

 > Algebraically, this is just the same as adding more constructors to
 `RuntimeRep`

 Well, yes, but

 * you'd have to double the amount of constructors in `RuntimeRep`
 (coherent and incoherent version of each).

 * more importantly, you couldn't be parametric in `RuntimeRep` and non-
 parametric in `Coherency` or vice versa.

 > This new design does not solve the problem I outline at the end of
 comment:47 though.

 Forgive my ignorance, but is it necessary to have the same `TYPE` in Core
 as is in surface syntax? Couldn't we just "forget" the coherency
 annotation everywhere and translate surface `TYPE :: Coherency ->
 RuntimeRep -> Type` to `TYPE :: RuntimeRep -> Type`?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11715#comment:54>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler