[Mailman-Users] Mailman reusing message-id, leads to duplicate message suppression

[Richard Damon]

On 11/5/13, 4:25 AM, Ralf Hildebrandt wrote:
> Hi!
>
> Mailman re-uses the message-id of the original email. In our case an
> exchange user is recipient of two mailing-lists. Both mails are
> adressed in the original message, both lists distribute the list, two
> emails having the same message-id arrive at the Exchange server, which
> seems to discard the second mail.
>
> Who's to blame? Mailman for re-using the message-id? Exchange for
> dropping the sencond mail on the floor?
>
This is really a tough problem. Presumably, since mailman doesn't
significantly alter the message (which is one reason it is allowed to
maintain the message-id) it should really matter that the recipient only
gets one copy of the message, or which one.

The source of the problem is that maintaining the message-id is
important for message references which are used for threading to work
well. People who are CC to the message, and use the copy they got
directly, and reply back to the list, will want there reply marked as a
reply to the list message. This REQUIRES that the message given to the
list have the same message-id as the incoming message.

The idea of dropping second copies of messages also can make some sense,
if you already have a message, do you really need a second copy? This is
almost absolutely right if the two copies are due to being listed twice
in the distribution, or the duplication due to automatic verbatim
forwarding from one address to another. It isn't quite so for sure in
the case of mailing list, or if the user wanted to compare things like
received headers to compare path ways. I suspect that the people
implementing this feature really didn't think about the mailing list case.

Note that another use of message-ids, in Usenet, has slightly different
rules, and there it is vital that duplicate message-ids be dropped due
to the fact that a multi-peered server should normally get many
duplicate message-ids from the various peers. But in Usenet, you don't
have the case of a central "mailing list" hub, gathering messages, being
allowed to make administrative changes, and then redistributing. I don't
think Microsoft is reusing Usenet code, but someone may have been
confused by Usenet RFCs, and applied them to email, whose system has a
lot of similarity (and even share some RFCs), but IS a different system.

In conclusion, the fault probably is on the mail system, but they can
probably claim they had good reasons to do it. Hopefully there is a way
to turn off the option if really needed.

This is also a problem with Google, which does the same thing, with no
option to turn it off. The problem more often shows up there by people
wonder why their messages don't make it to the list, because Google
removed the list distributed one, because the poster had a copy of the
message in their outbox.

-- 
Richard Damon