#11646: Make pattern synonym export type mismatch a warning -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 8.0.1-rc2 checker) | Keywords: Resolution: | PatternSynonyms Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | PatternSynonyms/AssociatingSynonyms| -------------------------------------+------------------------------------- Comment (by mpickering): You understand correctly. The check would be better but it is difficult to specify when a pattern synonym should and should not be allowed to bundled. Do you have an example where the check prevents you from writing the program you want to write? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11646#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#11646: Make pattern synonym export type mismatch a warning -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature request | Status: closed Priority: normal | Milestone: Component: Compiler (Type | Version: 8.0.1-rc2 checker) | Keywords: Resolution: wontfix | PatternSynonyms Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | PatternSynonyms/AssociatingSynonyms| -------------------------------------+------------------------------------- Comment (by dfeuer): Replying to [comment:3 thomie]:

No response from submitter. Unclear which problem this feature would solve. Please reopen if you disagree.

I disagree on philosophical grounds somewhat more than practical grounds. Pattern synonyms are fundamentally a ''syntactic'' feature. The essential idea is that we're writing introduction and elimination functions and tying them to construction and pattern matching syntax. In the associated pattern syntax, we tie a type constructor ''name'' to pattern synonym ''bindings''. The type checker has essentially nothing useful to contribute at the export stage. I think it should really stay completely out of the way. Others disagree, for their own reasons, but I'd like to at least be able to say that ''I'' don't care and I'd like it to leave me alone. I'll be opening another ticket shortly relating to a more practical problem with the way pattern synonyms are typed, to which the most obvious solution is essentially "Just let me do what I want, because it can't hurt type safety." Same philosophy here. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11646#comment:4 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#11646: Make pattern synonym export type mismatch a warning -------------------------------------+------------------------------------- Reporter: dfeuer | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 8.0.1-rc2 checker) | Keywords: Resolution: | PatternSynonyms Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | PatternSynonyms/AssociatingSynonyms| -------------------------------------+------------------------------------- Comment (by simonpj): So to be concrete, you want to allow this: {{{ module M( Maybe( Nothing, Just, P ) ) where pattern P x = [x, True] }}} So now an importing module can import `Maybe(..)` and will get two data constructors for `Maybe` and one for lists. I agree that there is nothing unsound about this, just a bit unexpected. I'd be perfectly OK with issuing a warning rather than an error. But doing anything at all requires ''some'' work from someone. And it may well be that someone then pops up to say "it really really should be an error". So a compelling use-case or two, and evidence of interest from other users would help. Does anyone else have an opinion. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11646#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler