Re: [GHC] #11228: Interaction between ORF and record pattern synonyms needs to be resolved.

24 Dec 2015

      #11228: Interaction between ORF and record pattern synonyms needs to be resolved.
-------------------------------------+-------------------------------------
        Reporter:  mpickering        |                Owner:
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  7.11
      Resolution:                    |             Keywords:
                                     |  PatternSynonyms, orf
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #9975, #11283     |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by adamgundry):

 * owner:  adamgundry =>
 * related:  #9975 => #9975, #11283

Comment:

 The panic was actually caused by the combination of pattern synonyms and
 `DisambiguateRecordFields` or `RecordWildCards`, without needing
 `DuplicateRecordFields`. See #11283.

 The more general question of the interaction of `DuplicateRecordFields`
 and record pattern synonyms looks like it will be rather intricate. At the
 moment, record pattern synonym constructors and fields are not
 distinguished by the `Avail` or `Parent` types, but constructors and
 fields will need to be handled differently. This will require significant
 refactoring.

 We need to track the pattern synonym to which a field belongs, so that we
 can disambiguate fields properly. This doesn't even require
 `DuplicateRecordFields`: merely using `DisambiguateRecordFields` fails
 when the constructor is a record pattern synonym. However, the pattern
 synonym isn't a "parent" for the field in the same way that a datatype is.
 And pattern synonyms have the added complexity of being able to be
 associated with arbitrary parent datatypes.

 In addition, once `DuplicateRecordFields` is involved we need to keep
 track of the `FieldLabelString` separately from the `Name` when
 representing fields (e.g. in `AvailTC` or `FldParent`).

 I'm not going to be able to address this in the near future.

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11228#comment:4
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler