On Tue, Dec 09, 1997 at 01:05:19PM +1100, Tyson Dowd wrote: cm >> Kai> If you can't get your mailer to reply to From: when you want to, cm >> Kai> complain to it's programmer - it's broken.
cm >> I thought that is the author sets reply-to, then that should cm >> be used for replies, and not from. I can reply to from: unless there cm >> is a reply-to, when that takes precedence. If people munge reply-to, cm >> I'll never knoe, will I? the rfc says that reply-to will be the address chosen automatically if the mua decides where to reply to. However, the rfc is very clear in stating that you should be able to send mail to any address in the header at your discretion (that you can choose to send replies to From or Sender instead. Errors go to Sender anyway, ignoring reply-to) sec 4.4.4 cm > From: [EMAIL PROTECTED] cm > Reply-To: [EMAIL PROTECTED] cm > cm >Becomes: cm > cm > From: [EMAIL PROTECTED] cm > Reply-To: debian-devel@lists.debian.org cm > Assuming no Sender line, or Sender = From, I beleive that the following mapping is compliant with the standard: >From -> Sender (Sender is omitted if it is the same as From, but it's not, anymore) Reply-To -> From (So From, as the rfc wants, shows the which machine the message came from) Reply-To: [EMAIL PROTECTED] In other words, ( reply-to == "debain-foo..." ? noop : (From == Sender || Sender == "") ? Sender = from, from = reply-to, reply-to = debain-foo : x-old-sender = sender, sender = from, from = reply-to, reply-to = debain-foo ) Now, the questions are, Can we do this? and Do we want to do this? Carl Mummert -- One must imagine Sysiphus happy. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .