2.54, 2.55

[luite]

The summary contains two separate test cases, GHC 7.4.1.20120406, looks like it's a problem with handling of some literals

# ghci
GHCi, version 7.4.1.20120406: http://www.haskell.org/ghc/  :? for help
Loading package ghc-prim ... linking ... done.
Loading package integer-gmp ... linking ... done.
Loading package base ... linking ... done.
Prelude> 2.55
*** Exception: divide by zero
Prelude> 2.54
<< gets stuck in infinite lopp >>

Lots of other literals also have this problem. The infinite looping version seems to get stuck determining the gcd, gdb backtrace gives this:

#0  0x000000010cebf042 in __gmpn_gcd_1 ()
#1  0x000000010ceb8d81 in integer_cmm_gcdIntzh ()

dtruss lots of these:

sigreturn(0x7FFF6E530220, 0x1E, 0x11104D5C0)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x7FFF6E530220)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x11104D5C0)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x7FFF6E530220)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x11104D5C0)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x7FFF6E530220)		 = 0 Err#-2
sigreturn(0x7FFF6E530220, 0x1E, 0x11104D5C0)		 = 0 Err#-2

The linux version doesn't seem to have this problem.

[simonpj]

Wierd. Paolo can you investiage please?

[pcapriotti]

Replying to simonpj:

Wierd. Paolo can you investiage please?

I can't reproduce on a 32 bit mac, and I don't have a 64 bit machine.

[simonmar]

Perhaps this is another symptom of #5810, which should now be fixed on the 7.4 branch. luite, could you check with the latest 7.4 branch?

[igloo]

I can't reproduce it with a validate build of HEAD on OS X amd64.

[luite]

I'll test it once I can get the 7.4 branch to build:

compiler/deSugar/Coverage.lhs:161:20:
    Not in scope: `getModificationTime'
make[1]: *** [compiler/stage1/build/Coverage.o] Error 1

[luite]

seems to be fixed now in the latest 7.4 branch, thanks!