hi,

i use a milter filter and some email lists in the aliases file.
the problem is that the milter scans before the aliases list expands
the members. but i want to scan with the milter for every single member
of the list.
does anyone have a solution to expand the list before the milter?

/thx,chris

On May 14, 10:16 am, "chris" <boes...@gmail.com> wrote:
> hi,
>
> i use a milter filter and some email lists in the aliases file.
> the problem is that the milter scans before the aliases list expands
> the members. but i want to scan with the milter for every single member
> of the list.
> does anyone have a solution to expand the list before the milter?
>
> /thx,chris

To obtain exapnded recipients from sendmail, the semi standard way is
to fork a sendmail -bv and parse the output

milter-rrres allows you to access sendmail -bv directly from the
milter api as well as calling rulesets and maps.

http://www.jmaimon.com/sendmail/

Depending on your setup, you can also parse the alias file or whatnot
yourself.

It occurs to me, that after the changes in 8.14 to eliminate/support
error recipient handling, sendmail might actually be able to supply
the expanded recipient list directly to the milter.

On 05/14/07 09:16, chris wrote:
> does anyone have a solution to expand the list before the milter?

This not (really) possible with current email / milter SMTP
implementations. More this is a limitation of SMTP, not milters. SMTP
sends an envelope containing one sender, one or more recipients, and a
message. Milters operate on the envelope, not individual recipients.

Now there are some things in development to enhance SMTP to offer per
recipient results / scanning during the SMTP transaction. However this
is not a reality as of yet.

The closest thing that can be assured to work is to limit recipients to
one per envelope so that you can be assured to have individualized
settings in each milter pass.



Grant. . . .

Grant Taylor <gtaylor@riverviewtech.net> wrote:
[...]
> The closest thing that can be assured to work is to limit recipients to
> one per envelope so that you can be assured to have individualized
> settings in each milter pass.

But one should be aware that this is a protocol violation as conforming
MTAs must be capable of processign at least 100 recipients in a single
SMTPÂtransaction. So the sending MTA would not need to retransmit but
could legally abort the whole message run and signal an error to the
sender.

Ralf
--
Ralf Döblitz * Schapenstraße 6 * 38104 Braunschweig * Germany
Phone: +49-531-2361223 Fax: +49-531-2361224 mailto:doeblitz@doeblitz.net
Homepage: http://www.escape.de/users/selene/
Mit UTF-8 kann man gleichzeitig äöüßÄÖÜæœłø¼½¾¤¹²³¢€£¥¶§ ¬Ã·Ã—±©®™¡¿ verwenden…

On 5/14/2007 11:29 AM, Ralf Döblitz wrote:
> But one should be aware that this is a protocol violation as conforming
> MTAs must be capable of processign at least 100 recipients in a single
> SMTPÂ transaction.

It has been my experience that systems are seldom completely RFC
compliant. In fact a good friend and colleague once told me that RFCs
are not best viewed as the bible of how to do things but rather a very
good recommendation on how to be best inter operable. If you keep this
in mind, you will find that a lot of things strive to be as inter
operable as possible, though few achieve RFC compliance. Each thing has
its own varying degree of compliance. So, technically this can be an
issue, but I think you will find other things far more likely to be a
problem than this.

> So the sending MTA would not need to retransmit but
> could legally abort the whole message run and signal an error to the
> sender.

This is an interesting view point, and one that I had not thought of
before. I usually think of things trying to get as much through at one
time as possible and only re-transmit what did not make it through. The
idea being get things to go the best that you can and then deal with the
rest after the fact. I.e. program how to deal with the failure and work
with the rest.



Grant. . . .

P.S. Sorry if this is a double post. My MUA is acting a bit weird over
a satellite connection.

Grant Taylor <gtaylor@riverviewtech.net> wrote:
[...]
>> So the sending MTA would not need to retransmit but
>> could legally abort the whole message run and signal an error to the
>> sender.
>
> This is an interesting view point, and one that I had not thought of
> before. I usually think of things trying to get as much through at one
> time as possible and only re-transmit what did not make it through. The
> idea being get things to go the best that you can and then deal with the
> rest after the fact. I.e. program how to deal with the failure and work
> with the rest.

Of course, butas it is legal to abort the run one should be prepared
that this could happen. Some large list processing sites are rumoured
not to retry mails that are undeliverable in the first try, so
everything not explicitly forbidden should be weighed when deciding
wether to violate the RFC a little bit or not, especially when dealing
with customers' mail.

Ralf (violating some other parts of 821/822 and accepting the risks involved)
--
Ralf Döblitz * Schapenstraße 6 * 38104 Braunschweig * Germany
Phone: +49-531-2361223 Fax: +49-531-2361224 mailto:doeblitz@doeblitz.net
Homepage: http://www.escape.de/users/selene/
Mit UTF-8 kann man gleichzeitig äöüßÄÖÜæœłø¼½¾¤¹²³¢€£¥¶§ ¬Ã·Ã—±©®™¡¿ verwenden…

On 05/15/07 11:33, Ralf Döblitz wrote:
> Ralf (violating some other parts of 821/822 and accepting the risks involved)

I think the most important thing you said was in your closing line "...
accepting the risks involved ...". You have made an informed decision.
There are many thing in this industry that can be done, some should
and some should not. So long as you are making an informed decision and
are willing to accept the risks there in, you are ahead of the curve.



Grant. . . .

On May 14, 10:26 am, "jmai...@ttec.com" <jmai...@ttec.com> wrote:
> On May 14, 10:16 am, "chris" <boes...@gmail.com> wrote:
>
> > hi,
>
> > i use a milter filter and some email lists in the aliases file.
> > the problem is that the milter scans before the aliases list expands
> > the members. but i want to scan with the milter for every single member
> > of the list.
> > does anyone have a solution to expand the list before the milter?
>
> > /thx,chris
>
> To obtain exapnded recipients from sendmail, the semi standard way is
> to fork a sendmail -bv and parse the output
>
> milter-rrres allows you to access sendmail -bv directly from the

err. -bv is still on the TODO list. -bt is what was exposed.