On Thu, 16 Dec 1999, Martin Buchholz <martin(a)xemacs.org> wrote:
>>>>> "Jan" == Jan Vroonhof
<vroonhof(a)math.ethz.ch> writes:
Jan> Martin Buchholz <martin(a)xemacs.org> writes:
>> following Lisp backtrace. It's very important to fix this as soon as
>> possible, and I am willing to help,
Jan> There is a fix, but you vetoed it :-). This is caused by the
Jan> Xaw3d/Xaw breakage. Compile with plain athena as a work around.
Hey, I didn't realize this was a bug fix! I was trying to avoid
introducing a DE-stabilizing change into XEmacs. How come this crash
hasn't always been a problem?
From what other people have said[1], creating a subclass of an Athena
widget with the flat Athena headers, then running against a 3d Athena
will cause you pain.
Daniel, I'm suddenly much more interested in helping with this,
since
I don't really like people being able to crash my XEmacs by sending me
email.
*grin* The next patch went through today[2]. It will refuse point blank to
build with mismatched Athena headers and widgets.
It also prefers to use something it *knows*[3] is right -- the headers
is `neXtaw/' for the neXtaw library and so on. It's pretty robust
against slight mismatches.
I'm not convinced yet that it's an Xaw/Xaw3d mixup yet,
though.
Well, I currently have the gutter turned off because it killed my
machine. Then I tried using W3 and the progress stuff killed me.
Hmmm. Heres a test for it. I just built with matching headers and shared
libs; lets try that some test again...
/time passes/
Well. Still here. The new progress bars look positively awful with the
neXtaw library, or possibly just at all. They didn't cope well at all
with this XEmacs of mine. Still, no crash.
Personally, I am now convinced that at least some of the lossage with
the new widgets is Athena foolishness related.
Daniel
Footnotes:
[1] I have seen things crash, put flat Xaw back and had them work, but
only because other people said this would work (and it did :)
[2] MsgId <87ln6vgr03.fsf(a)inanna.danann.net> in xemacs-patches
[3] Where /knows/ is defined as `if the user isn't deliberately trying
to make me suffer' -- you can probably fool the detection, but not
without actually trying.
--
The best work of artists in any age is the work of innocence
liberated by technical knowledge.
-- Nancy Hale