whitelisting mailing lists with dccproc

> From: Nicholas Piper <nick-dcc@nickpiper.co.uk>

> ...
> Ah; a very good idea. I use a .procmail script to invoke dccproc. I
> can just *not* pass the mail through dccproc based on procmail rules.
>
> Therefore I use procmail for identifying whitelisted emails, not
> /var/dcc/whiteclnt.

Yes, but don't you need to pass the dccproc result against procmail
rules to detect the thresholds?

Others have suggested that dccproc have dccm style rejection thresholds
and I keep thinking about them.  However, I can't see a way to use
them.  If dccproc had spam thresholds, what would it do with them?
How would dccproc reject mail?  It couldn't and closest it can get to
discarding mail is to emit nothing on stdout?  Wouldn't that cause grief?


> ...
> > If you use `dccproc | procmail`, can't procmail choose to ignore any
> > 'X-DCC.*many' matches if certain 'To: T-U-G@yahoogroups.com' patterns
> > are matched?
>
> I thought it was a bad idea to do this, because it means that the DCC
> servers will collect lots of mailing list checksums. For example, your
> reply to me has apparently been send to 17 other people who have not
> whitelisted this mailing list.
>
> Is it not better to avoid sending the checksums of known and valid
> bulk mails to the servers ?

I think you must assume the worst, which is also the best.  The DCC
counts are either accurate or overstate the "bulkness" of mail.  Even
if you don't report your solicited bulk mail to the DCC network you
use, other recipients of it will.
Unless you run the DCC so that it accepts reports only from certified
secret spam traps, you must assume all bulk mail will be detected
and so whitelist all solicited mail.

All of that means that it is a waste of effort to try to avoid
reporting the checksums of solicited bulk mail.


Vernon Schryver    vjs@rhyolite.com