#7939: RHS of associated type not kind-checked ------------------------------------------+--------------------------------- Reporter: goldfire | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.7 Resolution: | Keywords: Os: Unknown/Multiple | Architecture: Unknown/Multiple Failure: GHC accepts invalid program | Difficulty: Unknown Testcase: ghci/scripts/T7939 | Blockedby: Blocking: | Related: ------------------------------------------+--------------------------------- Comment(by simonpj): Well analysed. * The inconsistency seems to be deliberate; look at `TcTyClDecls.getFamDeclInitialKind`. But there is no explanation of why the difference, so I think we should abolish the difference until proved otherwise. * With `-XPolyKinds` we get kind polymorphism by default in data decls. Eg {{{ data T f a = MkT (f a) }}} we get `T :: (k -> *) -> k -> *`. * By the same token it is perhaps reasonable that {{{ type family F a :: * }}} gives `F :: k -> *` * But getting polykinded ''result'' kind by default (even with `-XPolyKinds`) seems odd to me: {{{ type family F a }}} Do we really want `F :: k1 -> k2`? My suggestion is that we adopt the current top-level story, namely that the result kind of a type family is `*` unless you explicitly use a kind variable. That's inconsistent with the argument story, but the examples you give above really are quite confusing Comments, anyone? The fix is easy either way. Simon -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/7939#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler