4:04 p.m.
>>>> "sb" == sb <sb(a)dod.no> writes:
sb> After "apt-get dist-upgrade" upgraded XEmacs on my debian
sb> system yesterday, Gnus in XEmacs started hanging when fetching
sb> headers. XEmacs is completely dead, and doesn't respond to
sb> `C-g'.
I don't understand why C-g doesn't help, but it is known that there's
a bug in the regexp code. You can work around it by using 21.4.15,
but that will break all regexps that use shy groups---they will fail
to match correctly. The cause of the bug is unknown at this point.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
5:17 p.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs
>>>> "Stephen J. Turnbull"
<stephen(a)xemacs.org>:
I don't understand why C-g doesn't help, but it is known
that
there's a bug in the regexp code. You can work around it by using
21.4.15,
Not really an option for me, since I don't build XEmacs myself. I
just apt-get it.
One thing I can do, is to use debian's non-mule version of XEmacs.
Appearently the problem isn't present there.
but that will break all regexps that use shy groups---they will fail
to match correctly. The cause of the bug is unknown at this point.
An article in posted to ding(a)gnus.org, mentions a fix applied to the
official freebsd port:
http://article.gmane.org/gmane.emacs.gnus.general/59500
I've no opinion on this patch's validity.
5:50 a.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs
>>>> "Steinar" == Steinar Bang <sb(a)dod.no>
writes:
Steinar> http://article.gmane.org/gmane.emacs.gnus.general/59500
That's basically the same as the reversion I mentioned. It's not a
fix, it's swapping one bug for another.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7:44 a.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs
>>>> "Stephen J. Turnbull"
<stephen(a)xemacs.org>:
>>>> "Steinar" == Steinar Bang <sb(a)dod.no>
writes:
Steinar> http://article.gmane.org/gmane.emacs.gnus.general/59500
That's basically the same as the reversion I mentioned. It's
not a
fix, it's swapping one bug for another.
Could you suggest a workaround in Gnus? A rewrite/restructuring of
the regular expressions, perhaps?
3:39 a.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs
>>>> "Steinar" == Steinar Bang <sb(a)dod.no>
writes:
Steinar> Could you suggest a workaround [for the XEmacs regexp
Steinar> infloop] in Gnus? A rewrite/restructuring of the regular
Steinar> expressions, perhaps?
No, not yet. The fact that removing the code that actually implements
shy groups short-circuits the infloop suggests a fencepost error in
that code, but I don't see it. Since it only happens in a specific
mode (message-mode) for me, I suspect the syntax table code (which is
known to be buggy, or at least incompatible with GNU Emacs's code).
See http://list-archive.xemacs.org/xemacs-beta/200412/msg00314.html
for the shortest test case I've been able to find, but note that it
only infloops in my current process, not in a fresh -vanilla invocation.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
12:25 a.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs [patch]
Lundi le 03 janvier 2005, vers 04:39:48 (CET), Stephen J. Turnbull a
écrit:
>>>>> "Steinar" == Steinar Bang
<sb(a)dod.no> writes:
Steinar> Could you suggest a workaround [for the XEmacs regexp
Steinar> infloop] in Gnus? A rewrite/restructuring of the regular
Steinar> expressions, perhaps?
No, not yet. The fact that removing the code that actually implements
shy groups short-circuits the infloop suggests a fencepost error in
that code, but I don't see it.
I think that I localized the bug. If I understand correctly, the loop
that is around line 4805 of regex.c should be written:
if (regs && regs->num_regs > 0)
for (mcnt = num_nonshy_regs; mcnt < regs->num_regs; mcnt++)
regs->start[mcnt] = regs->end[mcnt] = -1;
instead of:
if (regs && regs->num_regs > 0)
for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
regs->start[mcnt] = regs->end[mcnt] = -1;
Note the "num_nonshy_regs" instead of "num_regs" in the for
construct.
With this fix, I cannot reproduce the bug anymore, and it makes also
valgrind happy (no errors).
I attached a patch against version 1.25.2.11 of regex.c. The patch is
a bit long but it is mostly reformatting.
--
Arnaud
7:58 a.m.
New subject: [Bug: 21.4.16] Gnus CVS fails in xemacs [patch]
>>>> "Arnaud" == Arnaud Giersch
<arnaud.giersch(a)free.fr> writes:
Arnaud> I think that I localized the bug. If I understand
Arnaud> correctly, the loop that is around line 4805 of regex.c
Arnaud> should be written:
You sure do! That's it!
Thank you very much!
I'm going to add this ChangeLog and forward it to xemacs-patches:
2005-01-26 Arnaud Giersch <arnaud.giersch(a)free.fr>
* regex.c (re_match_2_internal):
Correctly initialize loop which clears uninitialized registers.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7239
days inactive
7266
days old
6 comments
3 participants
participants (3)
-
Arnaud Giersch -
Steinar Bang -
Stephen J. Turnbull