#16082: Sort out treatment of underscores in types -------------------------------------+------------------------------------- Reporter: goldfire | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.7 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: | -------------------------------------+------------------------------------- Changes (by adamgundry): * cc: adamgundry (added) Comment: A few of us discussed this today, and came up with the following: * Treat anonymous wildcards `_` just like user-supplied metavariables, for the "I don't care" case. GHC should try to infer a value for the metavariable, and does not report anything (unless there is a genuine error, i.e. it is not possible to either solve or generalise over the metavariable). In particular, it becomes possible to write anonymous wildcards in more places, such as datatype and instance declarations. * Report named wildcards `_w`, including both type/kind and their instantiation (if one is determined by the context/expression), for the "I want help figuring out what this is" case. * We can optionally add flags to control this behaviour, e.g. to print out the inferred values for anonymous wildcards or suppress the named ones. * Add a separate syntactic cue indicating a partial type signature (e.g. use `:?` instead of `::`) for case 3. This avoids the oddity that introducing a wildcard means suddenly the expression no longer has a complete type signature. This probably needs a GHC proposal as the next step. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/16082#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler