[GHC] #9701: GADTs not specialized properly

[GHC]

#9701: GADTs not specialized properly
-------------------------------------+-------------------------------------
              Reporter:  dfeuer      |            Owner:
                  Type:  bug         |           Status:  new
              Priority:  normal      |        Milestone:
             Component:  Compiler    |          Version:  7.9
            Resolution:              |         Keywords:
      Operating System:              |     Architecture:  Unknown/Multiple
  Unknown/Multiple                   |       Difficulty:  Unknown
       Type of failure:  Runtime     |       Blocked By:
  performance bug                    |  Related Tickets:
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------

Comment (by dfeuer):

 Replying to [comment:11 simonpj]:

 > Reverting to the main thread, I'm not sure where this discussion is
 going.  The base assumption is if multiple type-class instances match
 (e.g. from two enclosing pattern matches, comment:2, or from an enclosing
 patten-match and the top level) then you are on terribly thin ice is you
 want to predict which one will be chosen. They are all supposed to have
 the same value, and if they don't then, well, the ice is thin.

 Edward Kmett raised two concerns about the sort of specialization I was
 trying to accomplish: that it would break implicit parameters in some
 fashion and that it would break incoherent instances, I was attempting to
 allay those concerns, the latter by showing that the incoherent ice is so
 thin it's just not worth worrying about—there's so little exploitable
 predictability about it that we shouldn't worry about making it less
 predictable.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9701#comment:12>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler