[GHC] #4012: Compilation results are not deterministic

[GHC]

#4012: Compilation results are not deterministic
-------------------------------------+-------------------------------------
        Reporter:  kili              |                   Owner:
            Type:  bug               |                  Status:  new
        Priority:  high              |               Milestone:  7.12.1
       Component:  Compiler          |                 Version:  6.12.2
      Resolution:                    |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
 Type of failure:  Other             |  Unknown/Multiple
      Blocked By:                    |               Test Case:
 Related Tickets:                    |                Blocking:
                                     |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by simonmar):

 @thoughtpolice: did you check that `-fno-cse` works around it?  I thought
 it might be due to shortcutting rather than actual CSE.

 @Fuuzetsu we don't think there's a new regression in 7.10, but we think it
 is possibly more likely to happen due to the Applicative/Monad changes,
 because we often see `pure = return`.  This triggers the condition in the
 second bullet point in the description.

 I'm actively trying to convince people to work on this, because it's
 important. (I might end up working on it myself if I can find time).  But
 beyond temporary workarounds, it's very unlikely that anything will happen
 on the 7.10 branch.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/4012#comment:96>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler