Author: Phil Chambers Date: To: exim-users Subject: Re: [Exim] local_scan(), content-scanning, multiple recipients and bounces
On Thu, 6 Feb 2003 17:11:59 +0000 (GMT) Philip Hazel <ph10@???> wrote:
> On Thu, 6 Feb 2003, James P. Roberts wrote:
>
> > My own suggestion would be to explain to your users, at the time they
> > are choosing opt-in or opt-out, that this dilemma exists, and that,
> > should a message come in addressed to more than one user of your system,
> > and at least one of those recipients wants scanning to be done, then you
> > WILL scan.
>
> The only other option is to have 2 boxes (or two MTAs on the same box).
> The first one splits up messages into individual deliveries that have
> only one recipient, and delivers these to the second, which does the
> scanning.
>
But that means you accept all messages before putting them through local_scan().
A major point of local_scan(), surely, is that you can reject messages on the
basis of their content before the SMTP phase is complete.
This has always been my objection to most (probably all) current e-mail virus
scanning packages. They work by accepting everything, then scanning and trying to
reject later. That means you get masses of non-deliverable rejection messages on
your system. I work on the principle that I want to reject as much as I possibly
can at the SMTP stage. That transfers the onus from my MTA to the MTA which is
trying to send a problem message. It massively cuts down the number of deferred and
frozen messages on my queues.
Phil.
---------------------------------------
Phil Chambers (postmaster@???)
University of Exeter
