Excerpts from transarc.system.ietf-822: 22-May-01 Re: Mailing list
addition o.. Dave Crocker(_at_)brandenburg (2337*)
At 12:57 PM 5/22/2001, Keith Moore wrote:
With respect to mailing lists, I'll partially disagree with Keith. For
most mailing lists, the list agent is best thought of as a user, in terms
of MTA behavior.
treating lists as either "users" or "MTA" inevitably leads to confusion.
it's more realistic to say that they are somewhere in between, acting
as users at some times and as MTAs in others.
Hmmm. Thought my statements were rather more clear than your response
indicates:
With respect to MTAs, a sender and the list represent source and
destination points. Both are user agents, representing the beginning and
ending of an MTA sequence. The same for List and recipients, with the list
this time acting as source and recipients acting as, well, recipients.
Hence, a mailing list distribution causes two, discrete MTA
post-transfer-deliver sequences.
However there is a higher-level process at work, between the original
sender and the final recipients. At this level there is a single
transaction, covering the two list-mechanism subordinate transactions.
The difficulty with creating proper Reply behavior is that our standards
mechanism were not designed for this higher-level construct. It's a bit
like trying to represent 3 dimensions in 2.
What's the fundamental distinction between an agent redistributing mail
to a list and another agent forwarding mail to, say, a new email address
given in a .forward file? Is it only that the list redistribution
rewrites the Return-Path: whereas .forward processing doesn't? Is this
why you see list redistribution as more of a MUA function than an MTA
one?
Regards,
Craig