One might think that we wish to treat 'family' and 'role' as regular old varids whenever -XTypeFamilies and -XRoleAnnotations are off, respectively. But, there is no need to do so. These pseudo-keywords are not stolen syntax: they are only used after the keyword 'type' at the top-level, where varids are not allowed. Furthermore, checks further downstream (TcTyClsDecls) ensure that type families and role annotations are never declared without their extensions on. In fact, by unconditionally lexing these pseudo-keywords as special, we can get better error messages.
Also, note that these are included in the `varid` production in the
#10996: family is treated as keyword in types even without TypeFamilies enabled -------------------------------------+------------------------------------- Reporter: oerjan | Owner: Type: bug | Status: new Priority: low | Milestone: Component: Compiler | Version: 7.10.2 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: GHC rejects | Test Case: type Test valid program | family = family Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by oerjan): I note this quote in [https://raw.githubusercontent.com/ghc/ghc/master/compiler/parser/Lexer.x lexer.x]: parser --
a key detail to make all this work.
So clearly it was ''intended'' that they not be stolen, but something got messed up, perhaps when declaring types with `TypeOperators` were allowed. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10996#comment:7 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler