#15591: Inconsistent kind variable binder visibility between associated and non- associated type families -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.6.1 Component: Compiler | Version: 8.5 Resolution: | Keywords: | TypeApplications, TypeFamilies 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): For comment:4's `C`, I would expect {{{ C :: Type -> Constraint -- a is required, but it's not mentioned by name in the kind T2 :: forall (a :: Type) (f :: Type -> Type). f a -> Type }}} In `T2`'s kind, I've put `a` first, because it was already in scope. This parallels the fact that class variables are quantified before other variables in polymorphic methods. Both `a` and `f` should be `Specified`, as they are user-written. The various `Type`s in the kind of `T2` arise from the defaulting behavior of open type families (unchanged in this conversation). The specification of `Specified`/`Inferred` is very simple: if the user has written the variable name in the declaration, the variable is `Specified`. If not, it's `Inferred`. As for the patch: I would expect that the solution to this problem would be in the type-checker, not the renamer. As it stands, I'm pretty sure that the renamer identifies the `a` in these associated type definitions with the same unique as the `a` in the class. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15591#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler