On Sun, 9 Nov 2003, Andrzej Filip wrote: > >>It will make remote host sending message to sendmail & cyrus pair responsible > >>for generating the bounce message (in format beyond our control). > > > > Right -- this doesn't satisfy the definition or reject. > > But it satisfy definition of usefullness when dealing with "most likely" > spammers (IMHO).
I'm not going to tell you what to do with your particular setup, but breaking standards is not something we're particularly eager to do. > > I don't think this is something that really needs to be pushed out to the > > MTA. Though sieve-at-MTA-time may be interesting, its an unrelated > > problem. > > The map protocol can be used to implement at-RCTO_TO-time sieve > interpretation by cyrus (passing data required, getting back the reply. Except there isn't enough information *at* rcpt-to time to run Cyrus's sieve implementation. > BTW: Have anybody suggested implementing "cyrus milter" ? > [milter: external "mail filter" for sendmail] > > It would allow full integration with sendmail, rejecting at "RCPT TO:" stage > (invalid addresses in domains handled by cyrus) and at "final dot" (anti-spam > sieve for "most likely faked" sender address). > [Milter API library (as provided by sendmail.org) requires thread support.] This would strike me as something better served by doing a sendmail Sieve milter, and not something cyrus-specific (the socket map is more than adequate for RCPT TO time). Of course, if you get more than one recipient at a time, you're going to lose in this case too. -Rob -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456 Research Systems Programmer * /usr/contributed Gatekeeper
