[GHC] #11167: Fixity of field-deconstructors incorrect

[GHC]

#11167: Fixity of field-deconstructors incorrect
-------------------------------------+-------------------------------------
        Reporter:  hvr               |                Owner:  adamgundry
            Type:  bug               |               Status:  new
        Priority:  highest           |            Milestone:  8.0.1
       Component:  Compiler          |              Version:  7.11
      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 simonpj):

 * owner:  kanetw => adamgundry


Comment:

 I'm puzzled.  We only use `Ambiguous` to defer to the type checker if the
 occurrence is, well, ambiguous.  But here it isn't: there is only one
 `runContT` in scope.  Ah: it's because, as kanetw points out, there's a
 missing case in the `OpApp` case of `rnExpr`.  So we could fix that at
 least, and then this example would work.  Let's do that anyway.  That will
 fix the regression.

 Now we'd only have a problem if (a) we have `-XOverloadedRecordFields` and
 (b) the selector occurrence really was ambiguous.

 And then I suppose that the right thing to do is to fail (at least if the
 fixities differ) saying "Ambiguous fixity for runContT" or something like
 that.

 Stuff needs to be fixed in the user manual too, to explain this.


 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11167#comment:9>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler