Andy Piper <andy(a)xemacs.org> writes:
This is the bug that is hitting Michael as well. I don't
understand
why this can occur and without some sort of test case I'm not sure
how to debug it. I am assuming the domain is a window. If it is then
it shouldn't have become dead without the instance being removed
from the frame cache. It would be interesting to know what window it
is in relation to your gnus session.
I was hitting the same bug. I now disabled that particular check
and replaced it by
static void
check_frame_subwindow_instance_cache (struct frame* f)
{
Lisp_Object rest;
LIST_LOOP (rest, XWEAK_LIST_LIST (FRAME_SUBWINDOW_CACHE (f)))
{
Lisp_Object value = XCAR (rest);
ERROR_CHECK_IMAGE_INSTANCE (value);
}
}
int inhibit_subwindow_check = 0;
void
check_subwindow_instance_caches (void)
{
Lisp_Object frmcons, devcons, concons;
if (inhibit_subwindow_check)
return;
FRAME_LOOP_NO_BREAK (frmcons, devcons, concons)
{
check_frame_subwindow_instance_cache (XFRAME (XCAR (frmcons)));
}
}
I then stuck this test in Feval, Ffuncall & at the end of
Fset_window_configuration.
If then start gnus, open up the xemacs-review list in Gnus 5.8.7,
read a few message (some of them force gnus-picon to be loaded,
although no picons are displayed) and
q the Group I always get crashes in gnus-configure-windows.
The deleted window in question is the Gnus *Article* buffer.
Jan