#14332: Deriving clauses can have forall types -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 8.2.1 checker) | Resolution: | Keywords: deriving Operating System: Unknown/Multiple | Architecture: Type of failure: GHC accepts | Unknown/Multiple invalid program | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): OK, that's a start. But I have more questions: * Are the `forall`'d type variables in a `deriving` clause's type assumed to be distinct from the type variables bound by a data type itself? If so, how should GHC treat derived instances that would only work if kind unification were to occur? For instance, `data Proxy (a :: k) = Proxy deriving (Generic1 :: (k -> Type) -> Constraint)` is clearly OK, but `data Proxy (a :: k) = Proxy deriving (forall k. Generic1 :: (k -> Type) -> Constraint)` isn't, since it can't unify the `k` from the derived type and the `k` from the datatype. What should GHC do here? * What happens if one of the variables isn't quantified over (e.g., `data D = D deriving (forall a. C a b)`)? Is this an error? Many I'd be convinced if I did see a full write-up of this in a proposal. But as it stands, I'm quite unconvinced that this would work out if you actually sat down and attempted to implement this. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14332#comment:4 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler