#12088: Promote data family instance constructors -------------------------------------+------------------------------------- Reporter: alexvieth | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 8.1 checker) | Resolution: | Keywords: TypeInType Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: #11348 | Differential Rev(s): Phab:D2272 Wiki Page: | -------------------------------------+------------------------------------- Comment (by simonpj): I don't really understand your suggestion Richard. Firstly, we often infer the kind of a data type, so it's hard to separate dependence on its kind from dependendence on its definition. I din't undersand the precise dependency analysis that you informally propose. Secondly, I think that open type families are indeed the odd one out, as you acknowledge in the "odd conundrum". I don't see what's gained by separating types from defns even if we could. Thirdly, you suggest kind-checking all equations for a type family together as a single group. But you can't do that if they mention data types that have not yet been defined. Yet we can't delay too long ... getting those instances into play early is the whole point. Fourth, the dependence may be indirect. In the example, Open2 mentions Open1 directly. But it could instead mention F directly, where F is an open family that has an equation like `type instance F [t] = Open1 t`. Now the dependence on Open1 is hidden. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12088#comment:14 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler