#11650: Documentation does not mention that default definitions for Alternative(some, many) can easily blow up -------------------------------------+------------------------------------- Reporter: bgamari | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Documentation | Version: 7.10.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by RyanGlScott): Replying to [comment:8 ekmett]:
I mostly lack a preference here, but if we make it match the semantics of `Proxy` it'd avoid complications later on if we were to ever decide to make the leap and unify `(:*:)` with `Product`, `(:+:)` with `Sum`, `U1` with `Proxy`, etc. rather than retain two of each.
That's a pretty convincing argument. It does seem like it would save us work in the long run to make `U1` as `Proxy`-like as possible (and similarly with other `GHC.Generics` datatypes), so I'll do this before 8.0 lands. Some questions we should answer: 1. Do change `U1`'s derived `Eq`, `Ord`, `Read`, and `Show` instances (which have been around for a while) and manually implement them with lazy pattern-matching? That seems like the sensible thing to do, although it would be a subtle semantic change. 2. Do we add all of the extra instances which `Proxy` has but `U1` doesn't? (`Enum`, `Bounded`, `Ix`, etc.) My gut reaction is to punt on this and only add them if someone requests it. Plus, if we decide to merge `U1` and `Proxy`, it will be a moot point after that. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11650#comment:11 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler