From: Dennis Taylor <corbeau@execpc.com>
  To  : ggi-develop@eskimo.com
  Date: Thu, 18 Mar 1999 17:47:07 -0600

Re: giiClose() coredump

On Thu, Mar 18, 1999 at 10:24:43PM +0100, Marcus Sundberg wrote:

> Yep, it shouldn't make a difference. Btw just fyi, you don't have to
> use ggiFlush() in SYNC mode.

	Heh, good point. Force of habit, I guess.


> Judging by my mailings with Andy it's most likely a bug in the dynamic
> linker - I'm having no problems with neither Linux glibc 2.1 nor Digital
> Unix.

	Hmm. The box in question is a pretty stock RH 5.1 with all 5.2
patches, kernel 2.0.36, libc 2.0.7, GNU ld 2.9.1, gcc 2.7.2.3 on a
PII. I'll try it on a similar machine at work, if I have time, to see
if I can duplicate it.


> As a possible workaround you could try this diff.
> It will close the inputlib(s) before closing the target libs.
> (I have no idea if it makes a difference as can't reproduce the
> segfault.)

	Yep, this did the trick. With the patch applied, it works fine
for the demo apps and my little test program. Thanks a million!


__________________________________________________________________________
Dennis Taylor           "Anyone whose days are all the same and free from
corbeau@execpc.com       want inhabits eternity of a sort."  - Peter Hoeg
__________________________________________________________________________
   PGP Fingerprint: E8D6 9670 4FBD EEC3 6C6B  810B 2B30 E529 51BD 7B90