#8581: Add support for explicitly-bidirectional pattern synonyms -------------------------------------+------------------------------------- Reporter: cactus | Owner: Type: feature request | Status: new Priority: normal | Milestone: 8.0.1 Component: Compiler | Version: Resolution: | Keywords: | PatternSynonyms Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: 5144 | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by mpickering): OK, I'm coming back to this ticket 16 months later with a much better understanding of the conversation. Gergo's last comment (:19) is informative. I still think that this restriction should be removed in line with what Simon suggests (`CE => CP`). I can't add much to Gergo's worries about the implementation but 1. I hope this two stage type checking could be simplified, it's something I want to eventually look at. 2. One obvious stop-gap solution would be allow a user to provide a separate type signature for the builder. This gets around the whole "not being able to infer the type" problem. This ticket is probably a long way out of cache by now Simon but do you have any thoughts about my first bullet point? do you think it would be possible to refactor the pattern synonym type checking so the type of the builder can be inferred? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8581#comment:29 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler