[Mailman-Users] traceback while sending digests

sorry i haven't had time to be more responsive. i've got a lot of other
work on my hands at the moment, so i'll throw in my best educated
guess instead of a thorough analysis, and hope it helps.

First off, i believe that this problem is either system dependent or
showing itself through list configs that i don't have. i'm not seeing
this under linux or solaris2.5.1, and both installations have taken a
bit of a beating.

secondly, this problem probably results from calling
os.popen(<prg>, "w") where <prg> does this right off the bat:

if not os.fork():
# do the work
os._exit(0)


the way i understand os._exit(0) is that it should only be used for
child processes, not parent processes, when a fork() is done. Perhaps
there are some underlying things in the system or list configs which
cause the parent program to exit before the child is done reading
stdin.

at any rate, if those who are experiencing this problem could try
replacing the os._exit(0) in the function "main()" in scripts/deliver
with "sys.exit(0)" or some such, this *might* fix the problem.

I *think* that should be changed anyway, but could be mistaken.

that's my best educated guess.

if it don't help, i hope to be able to address more of this later this
week, but can't promise anything yet.

good luck,

scott