Re: connection level whitelist/blacklists anyone?

[Chuck Yerkes]

Quoting Dan Nelson (address@hidden):
> In the last episode (Mar 11), Sam Robertson said:
> > Would this be useful to anyone?  Has it already been proposed?
> 
> Are you looking for a modification to spamassassin to allow the
> connecting IP to be passed to it so you can have rules trigger off it? 
> I agree that would be nice.

It would be good to not check for spam on mail being emitted from
inside a core (yes, for ISPs, no for corps).  Those tags, when the
hit 3-4 on newsletters, are embarrassing and unwanted.  Yes, I can
use other boxes for outbound mail; no, it's "nicer" to have all
edge machines the same and use DNS to guide the mail with the other
machines as backup.

The "right" way would be for sendmail to have some
rules to guide milter usage like it does for a number
of other things.

Scheck_milter and Scheck_SpamAssassinFilter
(e.g. check_${whatever you have as an Input Filter Name}).

If these rules return a "dont" (like check_rcpt and family),
the milter never gets called.

As it stands, we have to write this over and over for each milter,
except the milters that are closed and commercial.  As much as I
love running AV milters on mail coming from boxes already running
AV software, it's the wrong way to do it.

So we write it over and over.
/me wonders about putting the rule into sendmail and running it,
but sendmail doesn't really have library-ized routines to do
these core functions.