RE: ghc 6.2 panic or program segfault, depending on flags

First thing to do is to use -dcore-lint in case that nails it.

If that doesn't report anything, use
         -dverbose-core2core -ddump-occur-anal -dppr-debug
-ddump-inlinings 
and send me all the (voluminous) output.

Simon

| -----Original Message-----
| From: glasgow-haskell-bugs-bounces@xxxxxxxxxxx
[mailto:glasgow-haskell-bugs-
| bounces@xxxxxxxxxxx] On Behalf Of Abraham Egnor
| Sent: 27 January 2004 23:09
| To: glasgow-haskell-bugs@xxxxxxxxxxx
| Subject: ghc 6.2 panic or program segfault, depending on flags
| 
| If I compile my program with -O or -O2, I get the following:
| 
| [abe@basil:~/local/tla/genome/src/test] $ ghc -O -package genome -o
test
| test.hs
| ghc-6.2: panic! (the `impossible' happened, GHC version 6.2):
|         simplCore/Simplify.lhs:1538: Non-exhaustive patterns in
function
| bind_args
| 
| However, if I add -dcore-lint or remove the -O, the program compiles
but
| segfaults mid-run.  This bug showed up when I added an "Eq a" context
to a
| typeclass definition in some code in the genome package (which itself
| compiles fine with or without optimizations).
| 
| In addition, if instead of compiling the genome code as a package, I
build
| it into the test program using ghc --make, the resulting test program
| compiles *and* runs without problems, even with -O.
| 
| Unfortunately, as with Isaac Jones, I working with Aetion internal
code
| and can't post it, but I'll happily do any follow-up investigation
people
| suggest.
| 
| Abe
| 
| _______________________________________________
| Glasgow-haskell-bugs mailing list
| Glasgow-haskell-bugs@xxxxxxxxxxx
| http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs