[1828] in Moira
Re: strange list behavior
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Wed Jul 25 02:56:14 2001
Message-Id: <200107250656.CAA03134@the-other-woman.mit.edu>
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Jeff Bigler <jcb@alum.MIT.EDU>
cc: bug-moira@MIT.EDU
In-reply-to: Your message of "Fri, 20 Jul 2001 19:27:28 EDT."
<200107202327.TAA04609@multics.mit.edu>
Date: Wed, 25 Jul 2001 02:56:12 -0400
Hi Jeff,
> which contains the following:
>
> athena% blanche mitspo-request
> LIST:jcb
> "Jeff Bigler" <jcb@alum.mit.edu>
>
> Mail sent to mitspo-request used to just send mail to jcb@alum, which is
> the correct behavior.
I've looked at the code that generates the aliases file and can't
figure out how this could have ever worked this way. :-)
> In July, I added USER:jcb to LIST:jcb. I have
> subsequently deleted USER:jcb from LIST:jcb. However, since that time,
> mail sent to mitspo-request sends a copy to jcb@alum (correct), and a
> second copy to USER:jcb, which doesn't go through alum (incorrect).
>
> It seems like adding USER:jcb to LIST:jcb flipped a bit somewhere that
> didn't get reset when I deleted USER:jcb from LIST:jcb, possibly because
> the username happens to be the same as the list name.
If we ignore my comment above, this would generally mean corruption in
the imembers table (which how moire flattens lists internally.) I
verified that the relevant lists contain exactly the correct rows in
the imembers table, so this isn't the problem.
The problem appears to be that the aliases file entry for
mitspo-request contains jcb, meaning the list jcb which contains no
email-able addresses, but this is getting superceeded by the user jcb.
Obviously, you want this superceeding to occur in general, because
otherwise mail to jcb@mit.edu would go to the members of the moira
list jcb (ie. no one.) Since the list jcb is set as 'not a mailing
list' in moira we do consider it a bug that alias for mitspo-request
contains jcb (rather than the members of the list jcb.) We even have
code to fix this bug, but it requires a lot of testing, coordination,
and some other work to prevent undesired consequences. Unfortunately,
we haven't managed to complete this work, nor can I guess when we
will.
In the mean time, I suspect that renaming the list jcb to something
else will correct the symptoms that you're seeing. If you decide to
try this, I'd appreciate it if you could test it and confirm my
belief, since it's based mostly on reading the code that generates the
aliases file.
Jonathon
