[Mailman-Developers] Repeating messages

[Matt Helsley]

Well IMO moderated messages should not be grouped with rejected messages.
They are very different things.

I think a simple database for each list would solve the problem. A table
with the following info:

hash, message id, expiration time, reference

When a message is received at the site, it's hash is computed and compared
to the database entries of each list. If the hash or message id match and
the expiration time has not passed then look at the "reference". If the
reference says the email was delivered (and not held) then this is a
duplicate: discard/hold (see below) the email. If the message was held
previously, discard the held copy and attempt to complete the delivery
(do not remove the table entry -- maybe update the expiration time and
reference. Also, it may get held again). Any hold or delivery inserts an
entry into the table. Rejection/discards do nothing to the table.

Of course any expired database entries could be removed either during the
match, or in a separate sweep (maybe triggered by cron? a
daemon?) or using both methods. In fact the separate sweep would be
necessary to bound the database size.

The next question is how would such a system handle cross-posting? A
mailman installation I help manage does have one list that feeds another
and this could be a problem with the above system: intentionally posting
duplicates to different lists wouldn't work. Maybe there should be an
option to hold duplicates for approval rather than discard them?

Does that sound reasonable? 

Cheers,
	-Matt Helsley

On Thu, 22 May 2003, Michael Yount wrote:

> There is at least one problem case:  someone posts a message from an
> address that isn't subscribed to the list and receives a notice that the
> message was moderated/rejected.  That person then posts the message
> again from the correct address, and again receives a notice, this time
> because the message is a duplicate.
> 
> Michael
> 
> On 22 May 15:33, Nick Arnett wrote:
> > FYI, Listserv prevents duplicate messages by checksumming every message and
> > rejecting those that appear to be identical within N minutes... in case
> > nobody has ever suggested that feature.
> > 
> > And as long as we're on that subject, they also stop some spam by sharing
> > the message data and temporarily (24 hours, IIRC) blocking any address that
> > sends the same message to N mail servers around the same time.
> > 
> 
> _______________________________________________
> Mailman-Developers mailing list
> Mailman-Developers at python.org
> http://mail.python.org/mailman/listinfo/mailman-developers
>