#14132: Report an error for a missing class instance before an error for type family instances of an associated type. -------------------------------------+------------------------------------- Reporter: duog | Owner: duog Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 8.2.1 checker) | 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 simonpj):
This means moving ReporterSpecs between report1 and report2 doesn't work as is
I don't agree. I believe you can move "non-tv-eqs", "homo-eqs" and "otehr-eqs"; they are not "truly insoluble". Ah: I see that `trulyInsoluble` is a poorly-named predicate. It is only applied to constraints in `wc_insol`, and they are things like `Int ~ Bool`, but NOT things like `a ~ Int` (which are no necessarily insoluble). It'd be much clearer if `trulyInsoluble` really did check for truly- insoluble constraints. Separately I'd been thinking that separating `wc_insol` from `wc_simple` is a waste of effort; it'd be better to combine them. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14132#comment:4 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler