Charles Lindsey wrote:
In <489058DA(_dot_)9070103(_at_)network-heretics(_dot_)com> Keith Moore
I think it's rather odd that people are arguing that a proposed
extension to the 822 (i.e. user-to-user) protocol can only be used by
what is arguably mail transport (or on the boundary between transport
and the recipient's user agent). It certainly doesn't belong in SMTP.
Nor does it belong in a body part, since it's clearly a message-level
Eh? That header has no meaning whatsoever for RFC 2821 transport. Its only
possible effects arise _after_ 'final delivery', at which point it is
sitting in some sort of 'mailbox' (be it IMAP, POP3, or Hector's grand
archive with alread-existing purging mechanisms). So the feature certainly
has to be in an RFC 2822 header, if it exists at all.
But the idea of a message store wasn't part of the original internet
mail architecture. So it's not clear to everyone where its
responsibilities lie. If the message store is (as I believe) a part of
the recipient's user agent then it should be neutral with respect to
storing messages except when given explicit instructions by the recipient.
Unfortunately the market has swallowed the idea that the "network" has
the right to interfere with traffic between sender and recipient, and
that furthermore it has the right to do so without any accountability.
Which makes it very hard to build reliable communications systems.