#10056: Inconsistent precedence of ~ -------------------------------------+------------------------------------- Reporter: crockeea | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.8.1 Component: Compiler | Version: 7.8.4 (Parser) | Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: GHC rejects | Unknown/Multiple valid program | Test Case: Blocked By: | Blocking: Related Tickets: #10059, #10431, | Differential Rev(s): Phab:D4876 #14316 | Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): When talking about `(~)`, it's important to distinguish between two different uses of it: 1. As the equality type operator. 2. As a laziness annotation in `-XStrict` (e.g., `data Foo a = MkFoo ~a`). If it weren't for usage (2), `(~)` would not need any special treatment at all in the parser. But alas, because of (2), we initially parse all uses of `(~)` as laziness annotations, and perform an additional pass right after parsing to determine which uses of `(~)` are for actually for (1), and which are for actually for (2). (Note that the problems I describe about `(->)` in comment:38 would still be relevant even if `(~)` had no special treatment in the parser. In other words, if we decided in the future to have some other type operator with a precedence of -2 or lower, than we'd have to figure out how to answer The `(->)` Question.) Once that's taken care of, the fixity of `(~)` is handled like any other type operator. I only opted to wire in the fixity of `(~)` in comment:38 since it's negative, and you can't assign a negative fixity through an `infix -2 ~` declaration. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10056#comment:40 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler