#8177: Roles for type families -------------------------------------+------------------------------------- Reporter: simonpj | Owner: goldfire Type: bug | Status: new Priority: normal | Milestone: ⊥ Component: Compiler | Version: 7.6.3 Resolution: | Keywords: 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: | -------------------------------------+------------------------------------- Changes (by RyanGlScott): * cc: RyanGlScott (added) Comment: goldfire, I do have a need for this in order to do "real work"... at some point in the future after #9123 is fixed :) My use case is laid out in [https://ghc.haskell.org/trac/ghc/ticket/8516#comment:7 this comment] in #8516. I'm interested in using higher-kinded roles to increase the number of datatypes for which you can derive `Generic1` instances by replacing `Functor` constraints with `Representational` constraints (where `Representational` is the typeclass proposed [https://ghc.haskell.org/trac/ghc/ticket/9123#comment:5 here]). Unfortunately, at the moment, my proposed plan would actually //decrease// that number, because there are datatypes defined in terms of type families for which you can declare `Functor` instances. However, type families always default to `nominal` roles for its type arguments, which means they can't be `Representational`! Ack! It definitely seems like //some// type families should be able to have `Representational` (or even `Phantom`!) roles for its arguments. A prime example is from [https://ghc.haskell.org/trac/ghc/ticket/8516#comment:7 this comment]: {{{#!hs type family Id a where Id a = a }}} I have a strong hunch that `a` should be have a `representational` role. Unfortunately, I have no way to inform GHC of this. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8177#comment:31 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler