[GHC] #15986: Poor error message source location reporting with unsaturated type family

[GHC]

#15986: Poor error message source location reporting with unsaturated type family
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.8.1
       Component:  Compiler          |              Version:  8.6.2
      Resolution:                    |             Keywords:  TypeFamilies,
                                     |  TypeErrors
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Poor/confusing    |  Unknown/Multiple
  error message                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by RyanGlScott):

 Now that I think about this some more, I'm less confident in my
 declaration that "I know GHC can do better here". That's because the two
 error messages that I pointed out in the original comments actually come
 from two entirely different compilation passes.

 The latter error message (`Expecting one more argument to ‘F’`) arises
 during typechecking, which works over source code—that is, where
 everything has a `SrcSpan` attached. The former error message (`The type
 family ‘F’ should have 1 argument`), however, arises //after//
 typechecking (during some //post hoc// validity checks in `TcValidity`)
 when we only have `Type`s floating around. The problem is that `Type`s
 don't have `SrcSpan`s attached to all of their subcomponents like `HsType
 GhcRn`s do, so it's harder to tell where to report errors during validity
 checking.

 Perhaps this is a sign that this bug can't reasonably be fixed until
 #15479 is implemented (i.e., when typechecking returns `HsType GhcTc`
 instead of `Type`). Unless someone can see a better path forward here?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15986#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler