[GHC] #11648: assertPprPanic, called at compiler/types/TyCoRep.hs:1932

[GHC]

#11648: assertPprPanic, called at compiler/types/TyCoRep.hs:1932
-------------------------------------+-------------------------------------
        Reporter:  thomie            |                Owner:  goldfire
            Type:  bug               |               Status:  new
        Priority:  highest           |            Milestone:  8.0.1
       Component:  Compiler          |              Version:  8.1
      Resolution:                    |             Keywords:  TypeInType
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 goldfire):

 I've been wondering if this dragon would bite. And now it has.

 Consider

 {{{
 data T (a :: Proxy k)
 }}}

 Does that have a CUSK? According to the CUSK rules for datatypes, it does
 indeed: all of the type variables have a kind signature. But there's a
 problem in that we don't know `k`'s type! So this kind signature isn't
 complete. Instead, we should need to say, e.g.,

 {{{
 data T :: forall (k :: *). Proxy k -> *
 }}}

 Note the explicit `:: *` on `k`'s binder, which is necessary for us to
 have completely specify the kind.

 However, in looking more closely at this case, I don't actually think this
 is the problem. A mistake in detecting CUSKness should affect type
 inference and whether or not we have principal kinds, etc., but it
 shouldn't produce the drastic failure that we're seeing above. Somehow, it
 seems that `quantifyTyVars` is making a mistake. I don't have the ability
 to test this at the moment (I'm in an intermediate state) but should get
 to this later this week. I propose to fix the immediate problem (which
 should be straightforward) and then to delay the debate about CUSKs for
 another time.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11648#comment:10>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler