Additional recipients within same lose mail when quarantined/released...

am.lists am.lists at gmail.com
Fri Mar 16 20:15:45 CET 2007


On 3/16/07, Matt Hampton <matt at coders.co.uk> wrote:
> am.lists wrote:
> > Do we think this is a bit of functionality that might be solved in the
> > elusive MailWatch 2.0 or is this issue rooted more deeply, e.g. at the
> > MS or PF layer?
>
> The problem is really pre MailWatch as has been discussed in previously.
>
> > It seems to me that MailWatch could log the message as
> > separate rows in the DB per "known user", referencing the same
> > quarantine file.
>
> I orginally did something similar - I split the recipients out into a
> separate table and then iterated through them.
>
> HOWEVER - the actual status of a message was still determined by the
> first recipient in the envelope.
>

This would be fine for me, as I'm not implementing per-user bayes or
per-user rules aside from black/white lists, which I may have to
consider.... hmm...

> The minor hit I took by splitting recipients (I use Sendmail) was almost
> completely mitigated by the SpamAssassin Cache.
>
> This also allows a unique message ID for each individial message recipient.
>
>
> > Just a thought. To bastardize this in to the current
> > system, I'd think about adding a separate routine that iterates
> > through the actual listed recipients, and adding the intended
> > recipient to a "deliver_this_copy_to" field (which would have to be
> > added).
>
>
> Yep that would work but not as neat or traceable as the split envelope
> method.
>
But, IIRC, the split envelope is still a problem because of the
original To: line of the message not being rewritten. Or does the
split envelope break that apart?

Thanks for your thoughts.


> matt

Angelo


More information about the MailScanner mailing list