segfault bug

Henry Cejtin henry@sourcelight.com
Wed, 21 Nov 2001 21:32:45 -0600


It would probably be a good idea, but isn't needed for this bug.  I re-compiled
with native false and get the same failure.  I turned on GC_EVERY_CHECK there
and it dies the same way, but it gets very far along.  Pretty strange.
I'm confused by the `used' field in the stack being so silly.  The program
shouldn't ever use much stack (as near as I can see) and yet the value in
the used member wasn't insane, just way way way too big.
Any suggestions?  Have any relevant bugs been fixed since 2001-10-06?  I could
just re-make mlton and try it on the new code, but it is clear that the bug
showing itself is a bit tricky so even if it would go away it wouldn't be a
very strong argument that it has been fixed.