#4426: Simplify the rules for implicit quantification -------------------------------------+------------------------------------- Reporter: simonpj | Owner: Type: feature | Status: new request | Milestone: 7.12.1 Priority: highest | Version: 6.12.3 Component: Compiler | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: | Related Tickets: #7880 None/Unknown | Test Case: | Blocking: | Differential Revisions: Phab:D211 | -------------------------------------+------------------------------------- Comment (by monoidal): I think we can make an exception and remove the warning flag in 7.12. There's no point in explicitly specifying it, and if someone needs to use `-fno-warn-context-quantification`, his code will break in 7.12 anyway. Richard: When I was implementing this almost a year ago, I downloaded the entire hackage archive and did {{{ grep -r "type.*=.* =>" --include=*s | grep -v "forall" }}} and manually filtered out correct uses. This found 46 errors in 26 packages. This number is an underestimation, since it does not count "data", multiline definitions or other situations affected by the change (TH type quotations or signatures such as `forall b. b -> (Ord a => a)`). Based on this, I decided to make it a warning first. I can investigate further if needed. Simon: We cannot use `Explicit` instead of `Qualified` because `Explicit` has precedence over `Implicit`, and valid code such as `f :: Ord a => a -> b -> a; f = const` would break. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/4426#comment:20 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler