I've been trying to ignore this discussion for a while. I hope to
return to that stance soon, but a few observations:
(1) As David has pointed out, there is a MTA-MTA WG. I don't especially
want to see this discussion there, but, as soon as things turn to "if
only MTAs did things differently, I'd suggest that the charter of this
WG is being pushed very hard in the direction of that one.
(2) There are installations who are getting very concerned about the
maximum cumulative amount of message someone can receive in a given
amount of time. For some parts of the extended mail internet, huge
logical messages provide avenues for denial-of-service attacks as well
as for forcing receivers to pay for being sent trash (sort of the
equivalent of junk fax, given the cost of keeping most fax machines in
paper). Now the opportunity for this sort of thing has alwyas been
there, but MIME creates a much greater opportunity to have it happen by
accident.
Curiously, if large messages are sent as a single chunk, servers
have ways to defend themselves by rejecting such tings. The proposed
SIZE verb may make the rejection process more efficient, but doesn't
change the possbility. But there is no defense that I'm aware of, at
least in transport, against being flooded by huge objects that are
automatically disassembled into message/partial by a sending system.
Please think about this folks: not everyone has fixed cost,
large-fraction-of-a-megabaud-of-better connections to machines with lots
of free disk space.
(3) The SMTP-EX group spent a lot of time worrying about the issue of
conversion and who/what gets to authorize it. I'd encourage people to
go back and look at that stuff, since the same ground seems to be able
to be traversed again.
--john