[GHC] #2940: Do CSE after CorePrep

[GHC]

#2940: Do CSE after CorePrep
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                Owner:  simonpj
            Type:  bug               |               Status:  new
        Priority:  lowest            |            Milestone:  8.0.1
       Component:  Compiler          |              Version:  6.10.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Runtime           |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by nomeata):

 I did? Oh dear. It must have been before I started to dislike CSE very
 much, because it gets in the way of rules, and can creasing sharing (e.g.
 by cse’ing `[0..1000]`). Or maybe it was precisely when I startd to
 dislike CSE... :-)

 Speaking about the phases CSE should happen: I very much believe we need a
 CSE phase as an Stg-to-Stg-transformation, to avoid reallocating `Left`s
 in the code for a `Either`’s `fmap`, for example. (There is a ticket for
 this, cannot find it right now.)

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/2940#comment:19>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler