[GHC] #3306: Improve syntax for GADT + records

[GHC]

#3306: Improve syntax for GADT + records
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                   Owner:  simonpj
            Type:  feature request   |                  Status:  closed
        Priority:  normal            |               Milestone:  6.12.1
       Component:  Compiler          |                 Version:  6.10.2
      Resolution:  fixed             |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:                    |  Differential Revisions:
-------------------------------------+-------------------------------------
Changes (by skvadrik):

 * failure:   => None/Unknown


Comment:

 6 years later: removed deprecated syntax for GADT constuctors.

 This allows to get rid of 2 reduce/reduce conflicts in parser. Conflicting
 productions were:


 {{{
 gadt_constr -> con_list '::' sigtype
 gadt_constr -> oqtycon '{' fielddecls '}' '::' sigtype
 }}}

 Recursive inlining of con_list and oqtycon helped reveal the conflict:

 {{{
 gadt_constr -> '(' CONSYM ')' '::' sigtype
 gadt_constr -> '(' CONSYM ')' '{' fielddecls '}' '::' sigtype
 }}}

 between two types of GADT constructors (second form stands for deprecated
 syntax). However, old syntax allowed `oqtycon` in constructor names, while
 new syntax allows `con` (which is not the same). There are therefore some
 constructor names that cannot be expressed in terms of new syntax (e.g.
 `(*)`).

 Note that ordinary data constructors are parsed in a different way than
 GADTs: types and names are mixed, so names have to be parsed as types and
 then converted to names. In GADTs constructor names and types are
 separated by `::`, so names can be parsed as just names and not as types.
 This results in a difference in the set of names allowed in GADT and
 ordinary constructors. See `Note [Parsing data constructors is hard]` and
 `Note [Difference in parsing GADT and data constructors]`

 Differential Revision: https://phabricator.haskell.org/D1118

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/3306#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler