[GHC] #13261: Consider moving Typeable evidence generation wholly back to solver

[GHC]

#13261: Consider moving Typeable evidence generation wholly back to solver
-------------------------------------+-------------------------------------
        Reporter:  bgamari           |                Owner:
            Type:  task              |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by goldfire):

 For the record, I've never been totally convinced by the "generate
 Typeable stuff at declaration time" approach. As noted above, there is a
 shared cost. Do we have any program that shows a concrete speed boost by
 the new design? Yes, the new design seems like `Typeable`-heavy code would
 run faster, but it would be lovely to verify this fact and be able to
 quantify it.

 Once there are dependent types, this issue does not really change. Yes,
 users could effectively use `Typeable` without writing the word `Typeable`
 in their programs, but they would still need to use runtime type tests to
 trigger any of this machinery. So I would consider this orthogonal.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13261#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler