[GHC] #9131: Experiment with a dedicated solver for Coercible

[GHC]

#9131: Experiment with a dedicated solver for Coercible
-------------------------------------+------------------------------------
        Reporter:  nomeata           |            Owner:
            Type:  task              |           Status:  new
        Priority:  low               |        Milestone:
       Component:  Compiler          |          Version:  7.8.2
      Resolution:                    |         Keywords:
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  None/Unknown      |       Difficulty:  Unknown
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+------------------------------------

Comment (by simonpj):

 We are mostly arguing terminology here.  Most solvers work by saying "let
 me take this goal, and solve it by breaking it up into sub-goals and
 solving those".  That's exactly what the `Coercible`-specific code does
 here.  You do not need to use that language of "instances" unless you want
 to.

 What the current solver does ''not'' do is ''search'', exploring many
 different paths to solving the goal.  And indeed search is problemantic
 when combined with the need to infer a substitution for unification
 variables.

 Anyway, we don't need to discuss terminology!  This ticket is really about
 whether a different solution strategy would be better.  And thus far I
 don't see any candidates on the table.

 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9131#comment:5>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler