#11142: Type-level skolem capture leads to core lint error -------------------------------------+------------------------------------- Reporter: goldfire | Owner: bgamari Type: bug | Status: new Priority: high | Milestone: 8.0.1 Component: Compiler (Type | Version: 7.10.2 checker) | Resolution: | Keywords: 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): That's true -- but we're not talking about kind generalization here, because the user is asking for it (albeit a bit indirectly). So: all kind variables that GHC invents are always quantified at the top level, but previously, users could specify kind variables to be quantified below top level. If that ability has been removed, this is a real loss of expressiveness that will bite. However, it's all easily recovered with the capabilities of my branch, where you can write kind quantification directly. So perhaps it all works out. But there's a bad migration story, because there will be no way to write higher-rank kind quantification that works in both GHC 7.10 and GHC 8.0. In any case, we need to clearly document the new behavior in the release notes. Is it worth trying to resurrect the old behavior on your branch, just to ease migration? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11142#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler