[GHC] #9132: takeWhile&C. still not fusible

[GHC]

#9132: takeWhile&C. still not fusible
-------------------------------------+-------------------------------------
        Reporter:  Blaisorblade      |            Owner:  skeuchel
            Type:  bug               |           Status:  patch
        Priority:  normal            |        Milestone:
       Component:  libraries/base    |          Version:  7.8.2
      Resolution:                    |         Keywords:  fusion
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  Runtime           |       Difficulty:  Moderate (less
  performance bug                    |  than a day)
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+-------------------------------------

Comment (by Blaisorblade):

 Thanks for taking care of this!

 The performance changes make sense to me, especially to fft2 which does
 use takeWhile. (I saw them on one OS X desktop, which however seemed to
 like suspending and had more stable timing; I didn't see them on a Linux
 machine which was otherwise more stable).

 On the regression, I agree it seems accidental. I'm confused that s1 and
 s2 are not joined by CSE. Delite/LMS, for instance, uses hash consing to
 prevent such programs from being formed in memory (the second definition
 is in a nested scope, but it seems this should not be a problem here).
 But I guess CSE in GHC would break programs using IO after the definition
 of IO is inlined.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9132#comment:16>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler