#9953: Pattern synonyms don't work with GADTs -------------------------------------+------------------------------------- Reporter: simonpj | Owner: cactus Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler (Type | Version: 7.10.1-rc1 checker) | Keywords: Resolution: | PatternSynonyms Operating System: Unknown/Multiple | Architecture: Type of failure: None/Unknown | Unknown/Multiple Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by simonpj): Well Richard, you are proposing that you can write `P1`'s type signature just like `T1`'s. Fair enough, I can see some merit; indeed it was my initial position. But then you '''must''' write `P2`'s type signature in some other way. What other way do you propose?? Since we already ''have'' two different ways to write the signature, we are at liberty to decide that those two different ways define two different behaviours; and that is what I am proposing. Yes, we could invent a third way, but that introduces a whole new raft of complications. (E.g. can you use that third way in a GADT declaration? For an ordinary function? etc) I'm totally open to considering alternative. But until we have one, let's get on with implementing the proposal we have. That is often illuminating; and if we come up with a better idea meanwhile we can retro- fit it easily enough. For data families, I think it should all work fine. But indeed we should check. Simon -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/9953#comment:13 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler