Several weeks ago I wrote:
XEmacs regularly hangs when doing networking. I use Gnus and read
news
from several news servers (and IMAP accounts). What often happens is
that XEmacs just hangs and stops responding if anything goes wrong with
a network connection. There seems to be no timeout, e.g. once it's hung,
it will stay that way (I've waited several hours). Nothing is redrawn,
there is just a blank XEmacs window.
The common recurring theme is the two messages that appear on the
console that XEmacs was started from:
X Error of failed request: BadWindow (invalid Window parameter)
Major opcode of failed request: 25 (X_SendEvent)
Resource id in failed request: 0xe009ba
Serial number of failed request: 593927
Current serial number in output stream: 594059
Xlib: unexpected async reply (sequence 0x9108b)!
xemacs-21.4.14: X Error of failed request: BadWindow (invalid Window parameter)
Major opcode of failed request: 18 (X_ChangeProperty)
Resource id in failed request: 0xe009ba
Serial number of failed request: 593926
Current serial number in output stream: 594059
The X requests are always the same, only the parameters change.
After being killed XEmacs sometimes produces a Lisp backtrace, which
usualy points (unsurprisingly) to network code, as in:
# (unwind-protect ...)
# (unwind-protect ...)
accept-process-output(#<network connection "nntpd<2>"
("nntp" . "news-server.san.rr.com") state:exit> 0 100)
# bind (process)
nnheader-accept-process-output(#<network connection "nntpd<2>"
("nntp" . "news-server.san.rr.com") state:exit>)
# (unwind-protect ...)
# bind (process)
nntp-accept-process-output(#<network connection "nntpd<2>"
("nntp" . "news-server.san.rr.com") state:exit>)
# (unwind-protect ...)
# bind (process wait-for buffer decode discard process timer pbuffer buffer)
nntp-open-connection(#<buffer " *nntpd*">)
[...]
As this is happening to me several times a day now, it is extremely
annoying. I seem to remember being able to keep XEmacs running for
weeks: has something changed, or have I started connecting to flaky
services?
And, most importantly: what can be done about it?
[umpteen or so crashes later, having lost some data along the way]
Can I do anything else to help resolving this bug? Is this a known
problem, or is it just me? Is anyone working on this?
I have a feeling this could be related to the resolver.
--J.