Hmm, no-one has any opinions on this issue one way or the other? If not, I'm inclined to use the string 'msg' instead of gethostname() and say that we don't support more than one machine delivering messages to TMDA at the same time.
> Doing some more thinking about how things would work if pending was a > true Maildir. > > Currently, the format of a pending message is: > > TIME.PID.msg. For example: > > 1038110763.80981.msg > > The confirmation address for this message looks like: > > [EMAIL PROTECTED] > > This allows TMDA to easily locate the corresponding message in pending > when a confirmation request arrives by parsing out the first two > portions of the cookie and appending '.msg'. > > However, the format of a true Maildir message in new/ is: > > TIME.PID.HOSTNAME. For example: > > 1038110763.80981.aguirre.la.mastaler.com > > I'd rather not imbed HOSTNAME within the confirmation address since > that would significantly increase its length. e.g, > > [EMAIL PROTECTED] > > So if I don't use HOSTNAME, then there is no reason to save the file > with HOSTNAME in the first place. I might as well continue to use the > string '.msg' instead. > > The ramifications of this would be that 1) it won't adhere exactly to > the Maildir spec. 2) You shouldn't have more than one machine > delivering messages to TMDA at the same time. It won't affect an MUAs > ability to read messages from the Maildir though. > > Since TMDA has never supported simultaneous deliveries from multiple > machines, and no one has ever requested it, I'm not sure it's worth > increasing the length of the confirmation addresses just to support > it. > > This is sort of a tough call though. Any opinions on this issue? _________________________________________________ tmda-workers mailing list ([EMAIL PROTECTED]) http://tmda.net/lists/listinfo/tmda-workers
