Crashes running emacs or xemacs: segmentation fault
Stephen J. Turnbull
stephen at xemacs.org
Thu Mar 10 20:26:35 EST 2005
>>>>> "Matthew" == Matthew Johnson <mjohnson at lightsurf.com> writes:
Matthew> May I suggest then, that in your replies, you set the
Matthew> Reply-To: header to point to the list, not to yourself?
If the reply-to header was set, please send me the mail headers so I
can see just what went wrong, and I'll fix the problem. I doubt it
was set, however; I think you probably just hit reply-to-sender. This
is a culture clash; I realize that very few Windows MUAs have
reply-to-list, and Windows users expect lists to set Reply-To.
However, we don't set Reply-To on our mailing lists; that is known to
cause problems in general, and would definitely cause problems for the
great majority of our list members, who do use RFC conforming MUAs and
do have frequent need to direct mail to individuals rather than back
to the list.
Matthew> I would also like to point out that in the text that said
Matthew> to reply to such-and-such address,
What text? In the post I sent? There's no reference to any address
in the body that I can find. In the crash error message? Yes,
crashes at xemacs.org is not very explicit, but it's clearly not a
personal name. In any case the post that you replied to was sent to
xemacs-beta, and that appears in the headers.
Matthew> Then I guess I have been lucky so far. I consider your
Matthew> response time very good.
You were lucky. I've been diligent lately. However, at the beginning of
February the spam buckets went two weeks without being checked.
>> Some of that information may be in the FAQ (see the Help menu,
>> XEmacs FAQ option, or the links to documentation on
>> http://www.xemacs.org/), however the FAQ is organized
>> differently from PROBLEMS.
Matthew> No doubt. But I notice that Q.2.0.6 of the FAQ also
Matthew> simply states without qualification that the PROBLEMS
Matthew> file "comes with XEmacs" without warning that it is only
Matthew> in the source distribution.
We'll fix the FAQ, then.
FYI, strictly speaking, we only try to support the source
distribution; 3rd party binaries (eg from Cywin and Linux distros) are
not our problem. Even the Windows binaries _we_ supply are a
3rd-party effort in some sense. There's no organizational commitment
to providing them. They are provided by one of the developers who
believes that Windows users by and large would not build XEmacs.
(N.B. He hasn't provided a binary for the last two versions.).
Matthew> And the only time 'segmentation fault' shows up (in a
Matthew> search at http://www.xemacs.org/) is when '__malloc_hook'
Matthew> can't be resolved.
Either the user already understands what "segmentation fault" means
and is capable of using the debugger to generate a back trace, or it's
all magic anyway, and they should just report the event as accurately
as they can. It's not the job of our documentation to give a course
in computer architecture and operation. This is not fixable in a
Matthew> So if, as you say, this error message is really being
Matthew> caused by Motif (more accurately, Lesstif) bugs, then
Matthew> that section of the FAQ needs to be corrected.
You've got to be kidding. Doesn't that go without saying, both
because what crashed is not a Lisp program, and especiall
More information about the XEmacs-Beta