#13408: Consider inferring a higher-rank kind for type synonyms -------------------------------------+------------------------------------- Reporter: goldfire | Owner: (none) Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Resolution: | Keywords: TypeInType 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): A few remarks about the code: - Put the `solveEqualities` around the first line, as the `tcLHsType` can emit constraints, and these will need to be solved. - Remove the second line. We don't need to run `kcTyClDecl` here. `kcTyClDecl`'s purpose in life is to generate constraints on the unification variables generates in `getInitialKind`. But we're not calling `getInitialKind`, so no need for `kcTyClDecl`. - Generalization looks about right. About the changed error messages: - The first will be hard to fix, for the reasons you describe. You could probably put some variant of `APromotionErr` in the env't to get a better message here. No, you can't just put the tycon in the env't, because you're making the tycon. Don't even think about `fixM` here. - The others just want for a `addTyConCtxt` around the call to `generalise`, which does a wee bit of validity checking. Otherwise, looks good! -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/13408#comment:8 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler