[GHC] #9964: GHC crash with NOINLINE and weird IO stuff

[GHC]

#9964: GHC crash with NOINLINE and weird IO stuff
-------------------------------------+-------------------------------------
        Reporter:  dfeuer            |                   Owner:
            Type:  bug               |                  Status:  merge
        Priority:  normal            |               Milestone:  7.10.2
       Component:  Compiler          |                 Version:  7.11
      Resolution:                    |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  crash                              |               Test Case:
      Blocked By:                    |  codeGen/should_compile/T9964
 Related Tickets:                    |                Blocking:
                                     |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by dfeuer):

 Replying to [comment:6 simonpj]:
 > Well that is indeed a bizarre example David.  Did it arise from real
 code in some way?
 >
 > Anyway I believe I've fixed it.
 >
 > Merge to 7.10
 >
 > Simon

 A while back, I asked whether `seq#` was really necessary to implement
 `evaluate`. I was trying to figure out if sufficient `NOINLINE` use would
 be able to avoid it. I was a bit surprised to crash the compiler! The
 example given is only slightly stripped down from my initial experiment.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9964#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler