#11498: GHC requires kind-polymorphic signatures on class head -------------------------------------+------------------------------------- Reporter: crockeea | Owner: (none) Type: bug | Status: closed Priority: normal | Milestone: Component: Compiler (Type | Version: 7.10.2-rc2 checker) | Resolution: duplicate | Keywords: CUSKs Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: #13365 | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by RyanGlScott): * status: new => closed * resolution: => duplicate * related: => #13365 Comment: OK, I now understand what is going on here. The fact that the second version of `C` doesn't typecheck in comment:2 is expected behavior. Once again, this all comes back to the fact that `C` doesn't have a complete user-specified kind (CUSK). As a result, the kind of `a` (in `class C a`) is chosen to be some fresh kind variable `k0`. When you try to declare a method of type `forall k. Proxy (a :: k) -> Proxy (b :: k)`, GHC complains because the kind `k` is bound in the method's type signature and //not// by the class (which binds `k0`), and thus `k` is not the same as `k0`. GHC perhaps ought to warn about the lack of a CUSK here, but that is the subject of #13365. Therefore, I'm opting to close this ticket and a duplicate of that one. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11498#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler