Martin Buchholz <martin(a)xemacs.org> writes:
Hi Martin.
I have such a system. No problems seen here with similar compiler
flags. Dynarr_resize is very stable code. I can only imagine a
compiler bug with double to integer conversions.
Hmm.
Are you using gcc 2.95.2, which I believe is the standard compiler
here?
Yes.
It's very hard to believe any change in 21.1 could have caused
your
backtrace. Nevertheless, you could try isolating the change, or try
changing `multiplier' in Dynarr_resize to be an int, and
unconditionally initialize it to 2.
Or try this patch, to avoid confusing the compiler:
[...]
The patch didn't apply (?). I changed the line by hand, but this
didn't help either, it still dumps core.
Next, I'll try to use integers.
norbert.