#15079: GHC HEAD regression: cannot instantiate higher-rank kind -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: high | Milestone: 8.6.1 Component: Compiler (Type | Version: 8.5 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: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): Richard and I discussed this yesterday. We noted that: 1. The interaction between higher-rank kinds and binder visibility has many parallels with other unexpected properties of higher-rank kinds. In particular, because there are (currently) no type-level lambdas, higher- rank kinds' `foralls` are more rigid than their type-level counterparts. As one example, in #13399 it was noted that higher-rank kinds' `foralls` don't "float" like the ones in higher-rank types. In this sense, it's perhaps somewhat understandable that higher-rank kinds' visibility might also be more rigid. 2. If we wanted to have a more permissive treatment of invisible binders in higher-rank kinds, then we'd need to come up with a better story. Richard considered various ideas like having a subtyping relation between visibilities, but in the end, he concluded that anything we could think of was likely far too complicated for such little payoff. Thus, the conclusion we reached was that we should keep the current behavior, and make note of this behavior in the users' guide. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15079#comment:7 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler