APPROVE 21.5 RECOMMEND 21.4
Thanks Hans!
"Vin Shelton <acs(a)xemacs.org>" <ethersoft(a)gmail.com> writes:
Martin, Stephen et al,
I don't know anything about 64-bit systems, but shouldn't size_t
really work correctly in the parsesndfile() call?
"(size_t) rrtn" would work, by converting to something with the right
size and the same value (assuming it's positive; size_t is always at
least as big as an int), but the original "(size_t *) &rrtn" formula
says "access storage of width sizeof(int) as if it were sizeof(size_t)
wide". IIRC on most 64-bit machines int is still 32 bits, while
size_t is 64 bits. (At least that's true with GCC on AMD64.) So you
get garbage in the other half of the value.
In any case, this patch apparently works around a real bug. Please
comment.
Well, given Martin's modesty, I'm sure unwilling to claim to really
understand the code, but I'm willing to bet future debugging time that
the patch is correct. So ... fire when ready!
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta