Thanks for your response Joe. We decided in the end not to chain the
milters and instead run the second milter on another downstream
sendmail box. The milter-rrres patch does look interesting, though.
I'll try taking a look at it in my spare time.

-Kai


jmaimon@ttec.com wrote:
> Kai Sung wrote:
> > Hi,
> >
> > I have two milters chained together. Is there a way for the second
> > milter to detect that the first milter deleted an envelope recipient?
> > I know I could do it by having the first milter add an X-header and the
> > second milter can detect the added X-header, but unfortunately I don't
> > have control over the first milter and it can't add X-headers. Any
> > other creative ideas would be appreciated.
> >
> > Thanks,
> > -Kai
>
> If you have a milter that you dont control, you may want to look into
> the milter-rrres patch.
>
> http://www.jmaimon.com/sendmail
>
> While the current posted version does not contain any post milter
> action ruleset hooks, the version I have under development does,
> untested as it may be. Perhaps you may want a crack at it. If you do,
> let me know and I will send you an interim.
>
> That may serve very well for you.
>
> You could perhaps monitor syslog for the message logged by these lines
> in sendmail/milter.c
>
> if (MilterLogLevel > 8)
> sm_syslog(LOG_INFO, e->e_id, "Milter delete: rcpt %s",
> response);
>
>
> You could also try modifying the functions called to update/set a macro
> when this occurs.
>
> More importantly, milters calls are done in order for every point of
> the smtp phase.
>
> You wont have one milter executing the xxfi_envrcpt callback after one
> has already called a smfi_delrcpt() inside of the xxfi_eom callback.