#14331: Overzealous free-floating kind check causes deriving clause to be rejected -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: merge Priority: normal | Milestone: 8.2.2 Component: Compiler (Type | Version: 8.2.1 checker) | Resolution: | Keywords: deriving Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: GHC rejects | Test Case: valid program | deriving/should_compile/T14331 Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): Huh? That is not what I would have expected at all. I would have expected: {{{#!hs class C {k1 k2} (a :: k1) (b :: k2) data D {k3} (r :: k3) = D instance forall k1 (a :: k1). C a (D a) where ... }}} In other words, you should unify `r` with `a`. After all, the `a` in `data D a = D deriving (C (a :: k))` scopes over the data type, not the `deriving` clause! Now if you had chosen to use a different scoping with `data D a = D deriving (forall (a :: k). C a)`, //then// I could see the instance being derived that you suggested. But I don't think users would expect that behavior to be the default (that is, in lieu of explicit `forall`s on a `deriving` clause, one should assume that the user-written type variables are bound by the data type if they appear in the `<tvbs>` in `data D <tvbs>`). -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14331#comment:14 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler