[GHC] #13549: GHC 8.2.1's typechecker rejects code generated by singletons that 8.0 accepts

[GHC]

#13549: GHC 8.2.1's typechecker rejects code generated by singletons that 8.0
accepts
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.2.1
       Component:  Compiler (Type    |              Version:  8.1
  checker)                           |
      Resolution:                    |             Keywords:  TypeInType
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by goldfire):

 I still think this is #13555. In all cases, the extra kind signatures
 mention kind variables, which GHC is now not generalizing.

 What surprises me about this case is that no generalization should be
 necessary to compile the singled `permutations`. My best guess is that
 untouchable type variables induced by some GADT pattern matching prevent
 GHC from inferring the correct kinds of the functions. If their kinds are
 generalized, then all works out. But since they are not being generalized
 any more, we run into a problem.

 I'm inclined to try to understand this as an infelicity in singletons,
 that it produces code that GHC can't infer the kinds of. I doubt
 singletons can fix the problem, but it would be nice to be able to
 characterize what exactly is the cause of the problem and then document
 the infelicity. In the meantime, I'd bet that adding a type annotation to
 `perms` and `interleave` in the `Data.Singletons.Prelude.List`
 implementation would allow singletons to compile. I'm happy to do the
 further digging, but not for a few weeks, I'm afraid.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13549#comment:11>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler