On Tue, 2006-06-27 at 03:11, jdow wrote: > > > > Rejecting the mail during the *SMTP* transaction *never* involves any > > hosts or addresses mentioned in the message headers. It is a TCP > > protocol-level thing only involving the peers: the sending host and > > your receiving host. It's impossible to involve a third party. > > > > Of course, that was the point Paul was making. > > That is, of course, the right way to do it. But after being on the > receiving end of a joe-job in the past I am a little "sensitive" to > the issue. And SOME people, probably not Paul on second thought (sorry > it was not first thought, Paul), are a little careless with regards > to "reject" and "bounce". If the mail has already been forwarded through a normal relay, it doesn't matter whether the next hop accepts and generates a bounce or it rejects with a 5xx status. The rejection will force any standards-conforming mailer to generate a bounce back to the sender - even if the sender address was forged. The only difference is that the reject saves your own machine the trouble of constructing and trying to return the bounce message. A lot of spam-spewing software sends directly though and a rejection is the end of it. -- Les Mikesell lesmikesell@xxxxxxxxx
