[GHC] #9872: Runing type functions is too slow

[GHC]

#9872: Runing type functions is too slow
-------------------------------------+-------------------------------------
              Reporter:  simonpj     |            Owner:
                  Type:  bug         |           Status:  closed
              Priority:  normal      |        Milestone:
             Component:  Compiler    |          Version:  7.8.3
            Resolution:  fixed       |         Keywords:
      Operating System:              |     Architecture:  Unknown/Multiple
  Unknown/Multiple                   |       Difficulty:  Unknown
       Type of failure:              |       Blocked By:
  None/Unknown                       |  Related Tickets:
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------

Comment (by goldfire):

 In thinking about this issue, the Right Way to get type families to be
 faster is to optimize them properly. By this, I mean taking the type
 family definitions and performing an optimization pass on them during
 type-checking/desugaring. We've essentially just implemented a tiny
 interpreter inside of !TcFlatten, and the whole thing could be more
 principled in approach.

 I'm not making a concrete proposal, just musing on the fact that we have
 here a classic problem -- how to write a fast interpreter for a given
 programming language. It just happens to be the language of type families.
 I suppose there is a body of research and experience on this very issue,
 and if we want to be serious about fast compilation times in the presence
 of heavy type-level computation, it would do well to use that body of
 knowledge.

 In any case, I'm done dwelling on performance for a while, but I'm quite
 pleased with my end result.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9872#comment:25>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler