[milters] Archive

Lists Index Date Thread Search

Article: 1120
From:
Date: 2006-08-25 18:03:49 -0400
Subject: Re: milter-spamc greylist if not in AWL

Removal...........: milters-request@milter.info?subject=remove
More information..: http://www.milter.info/#Support
--------------------------------------------------------

Quoting Anthony Howe <achowe@snert.com>:


> > What do you think about this? It gives the correct envelope return path
> > to spamassassin, which allows some SpamAssassin SPF rules to work that
> > would otherwise never fire.
>
> Its certainly feasible, but if you want to reject on SPF, its more
> efficient to do it pre-DATA with milter-spiff. Or use milter-spiff to
> tag the message and add the received-spf headers and then have SA pick

I certainly don't want to reject on SPF, just score somewhat. Too many large
sites have published SPF records that are just wrong (excluding some of their
main relays), never mind the forwarding problem, which is big for .edu's,
both incoming and outgoing.

milter-spiff would almost certainly be tighter C code... but then
milter-spiff's DNS queries and SpamAssassin's various RBL queries will be
serialized, while SpamAssassin makes attempts to run queries in parallel.
OTOH, milter-spiff would do the possibly long-running work before the DATA
phase. Could be better or worse depending on the situation.

I'm going to stick with SPF in SA. That way I can use all the default SPF
rules, and not have to overload them with my own versions based on a
milter-spiff header. There's CPU efficiency, memory efficiency, time
efficiency, and finally programmer brain cycles efficiency to consider.

Lists Index Date Thread Search