> I've tried to recreate a problem with this patch and have successfully
 > failed.  I'll continue using it for a few days, and I will let you know
 > if I see any problems, but I think that protecting window_list() from
 > quit fixes the problem(s) I reported.

Your impatience revealed this bug and your patience located it.  I now
attach a clean patch which also protects the occurrence in window_list_1
although that one doesn't appear so vulnerable - likely because quitting
always eats the maybe_quit from window_list first.  If you don't see any
problems with it within a week or so, I'll install it.

martin