Yes, in Bounce the reverse path should become null (MailImpl.setSender(null)). I had completely misunderstood the rules for bouncing.
We need a new getReversePath/setReversePath/<reversePath> and, for consistency, a new getFrom/setFrom/<from> (your suggestion yesterday) in the AbstractRedirect hierarchy, and modify the other getX/setX/<x> accordingly. The redirect mailet should have a "smart" default behaviour for the four "sender" related parameters (<returnPath>, <reversePath>, <sender> and <from>), as it has (and must continue to have) for <recipients> and <to>. My misunderstanding for Bounce came out from watching the existing "bounce" code in james: 1) MailImpl.bounce sets the reverse-path to the bounced mail recipients , and uses the Return-Path header set by MimeMessage.reply, that I don't know if set to NULL. 2) James.bounce sets correctly the Return-Path header to NULL ("<>") but sets the reverse-path to the "bouncer", that is also used for the new >From header. Both should be fixed (or not?). Vincenzo > -----Original Message----- > From: Hontvari Jozsef [mailto:[EMAIL PROTECTED] > Sent: martedi 1 luglio 2003 16.24 > To: James Developers List > Subject: Bounce, outgoing reverse path and fromAddress > > > If I understand well, the Bounce mailet should be modified to set the > reverse path of the outgoing mail to NULL. Currently it only sets the > Return-path header. > > On the other hand, IMHO, setting the Return-path unnecessary, that is the > task of the receiving server. > > Analysis: > Bounce currently sets up the outgoing notification message in this way: > -Return-path header = NULL > -envelope sender alias reverse path (and fromAddress) = "sender" > configuration parameter (assume postmaster) > -recipient: reverse path of the original incoming mail. (Actually it uses > the Return-path header of the original mail, but as I see James > always sets > that to the reverse path when smtp receives them message, so reverse path > and return-path are always the same from incoming messages.) > > Let assume that the other mail server is also a James server. What happens > is that: > -it receives the mail with a reverse path postmaster > -it overwrites the Return-path header of the mail with the reverse path, > i.e. our postmaster! > -the bounce has an invalid address (virus or spam) > -it takes the new Return-path, which is postmaster, and sends back the > message to out postmaster. Of courese this is what we wanted to avoid. > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]