Re: border display glitch

>>>>> "SJT" == Stephen J Turnbull <stephen(a)xemacs.org&gt; writes: SJT> I been taking your name in vain.... Barry is seeing the bug SJT> described at SJT> http://list-archive.xemacs.org/xemacs-beta/200205/msg00189.html. SJT> Any progress? For me, just in getting more confuzzled. :( >>>>> "BAW" == bwarsaw <bwarsaw(a)python.org&gt; writes: BAW> Building w/ a from-source fresh build of gcc 3.1 makes no BAW> difference. :( So I just brought up another box on Redhat 7.3, updated all my installed packages, rsync'd my XEmacs 21.4.8 source from my RH6.1 box, did a "make distclean ; ./configure --with-mule ; make" and guess what? No display glitches. I also tried the XEmacs 21.4.6 package that comes w/ RH7.3 and it also displays just fine. I'm baffled. The only difference I can tell between the two RH7.3 machines is that the first one (the one with the display glitch) I didn't install, while I did install the one that works fine. I can't see how that could make any difference though -- both are up2date with all packages. I actually can't get into the buggy box right now as that's at the office and the network appears down. :( Here at home, I'm using the following packages: XFree86-4.2.0-8 gcc-2.96-10 glibc-2.2.5-36 SJT> If I read Nix's comments right, it's GCC 3.x that is SJT> (possibly) doing something bogus, and I suspect RH has SJT> backported it. Possibilities: SJT> (1) --debug, --pdump, --use-union-type all change the layout SJT> of the code in ways that change optimizer behavior. SJT> Especially --use-union-type. Those options aren't advertised as available on the gcc on my home RH7.3 box. SJT> (2) Back up to vanilla 2.95.4. (I am _not_ seeing this on SJT> Debian.) SJT> (3) Ping Nix Who Seems To Know. (done with this message) Nix? What can you tell us? -Barry