Vin -- I'll see what I can do, in another week or so. It would help a lot if there were an easier way of reproducing the problem (Stephen?) -- since it doesn't occur with --vanilla i may not be able to reproduce it at all.
Stephen, can you also post some stack traces showing where it's looping? I can work on separating out the syntax patch but i'd like to see some of the traces so i can be reasonably convinced that this is indeed the location of the problem.
------- Original Message -------
On Sun, 09 Jan 2005 21:25:08 -0500 Vin Shelton wrote:"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
> (message "Prepare to infloop! Arming ...")
> (string-match "^\\([ ]*\\(\\w\\|[-_.]\\)+>+\\)+"
> ;; not executed if must quit from string-match == infloop
> (message "That was a rather short infinity!"))
> infloops for me in my current process, but not if I take out the call
> to message-mode, and (unfortunately) not at all in a fresh invocation
> with -vanilla. I suspect a problem with the syntax table, but that's
> as far as I'm going to get today.
> I would appreciate it if people would let me know what results are for
> a fresh invocation -vanilla, and also if Gnus users with a well-used
> Gnus invocation would see if that gives them an infloop. Also in
> Other suggestions?
You and/or Jan Vroonhof fixed a bunch of bugs in the syntax table code
in spring of 2002. That patch was never applied to 21.4, because it
was part of one of your mega-patches (message id
adrtpho9.fsf(a)xemacs.org) and it was a little too difficult to tease
out of the message. Can you please investigate this gnus infloop
problem in 21.4? Stephen and others think the problem may be related
to the syntax table - I'd really like to robustify that in 21.4, and
I'd even be potentially willing to risk temporarily de-stabilizing the
21.4 branch in order to get that ultimate robustness.
Please let me know your thoughts on this.