There's an engineering issue too. Richard is about to eliminate `RetTv`. But `RetTv` is essential to the implementation that accepts the above program. Are we going to leave `RetTv` for this sole (ill-defined)
#11431: GHC instantiates levity-polymorphic type variables with foralls -------------------------------------+------------------------------------- Reporter: gridaphobe | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.1 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): Replying to [comment:2 simonpj]: purpose? Yes, I recognize that as an unfortunate consequence of this suggestion. But it wouldn't be `ReturnTv` -- it would go back to the old `PolyTv`, just a tyvar that's allowed to unify with a polytype. Agreed that, if this works, it's a coincidence. That does seem to suggest that it doesn't work. My hope that it would work is because levity polymorphism is vanishingly rare. In any case, I don't feel strongly. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11431#comment:3 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler