dccm, sendmail, and fetchmail.

I'm bringing this spur back onto the list, per VJS's suggestion.


* On 2002.06.14, in <200206141712.g5EHCSpn025224@calcite.rhyolite.com>,
*	"Vernon Schryver" <vjs@calcite.rhyolite.com> wrote:
> > Fetchmail is a POP3/IMAP client, but it doesn't do any delivery of its
> > own. Rather it either passes incoming messages to a local MDA (procmail
> > is the common choice, though mail.local or anything else will work), or
> > to a local MTA (the default, and the only option originally supported).
> >
> > Probably, in this scenario, fetchmail is trying to "deliver" mail via
> > SMTP injection on localhost:25. Raymond originally took the attitude
> > that SMTP injection was the only way to assure proper processing of
> > inbound mail, but didn't account for the thousands of systems that don't
> > actually want to run MTAs -- this is why there's now an MDA option.
> 
> That suggests two things to me:
> 
>   - the right way for darkmark@filament.org to procede might be to
>      use dccproc inside procmail because sendmail is not involved
>      on the fetchmail client

Probably -- it at least avoids the whole problem with sendmail.


>   - if sendmail is involved, then more logging is needed to see why
>      sendmail is not invoking the milter interface.
> 
>     Sendmail only invokes milter filters when SMTP is involved.
>     Does fetchmail inject mail locally with `sendmail -bs` or with
>     with `sendmail`?  

Actual socket connection to localhost on port 25.


> Since the main information is your private note, you're more than
> welcome to the honor of sending to the DCC mailing list or
> darkmark@filament.org.

-- 
 -D.	dgc@uchicago.edu	NSIT	University of Chicago