#15497: Coercion Quantification -------------------------------------+------------------------------------- Reporter: ningning | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D5054 Wiki Page: | https://ghc.haskell.org/trac/ghc/wiki/DependentHaskell/Phase2| -------------------------------------+------------------------------------- Changes (by simonpj): * owner: ningning => (none) * status: closed => new * resolution: fixed => Comment: I came across `Note [Emitting the residual implication in simplifyInfer]` in `TcSimplify`. It says “We can't form types like (forall co. blah), so we can't generalise over the coercion variable, …” But when you commit coercion quantification, we ''can'' quantify over coercions in types. So it is possible that this bit of cruft could be tidied up, with a significant simplification. I'll re-open the ticket to keep this idea on the radar. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15497#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler