Markus Stumpf <maex-lists-email-ietf-smtp(_at_)Space(_dot_)Net> wrote:
On Mon, May 23, 2005 at 09:47:52AM -0400, John Leslie wrote:
Misdirected bounces are a significant problem:
Can we please stop using the term "misdirected bounces" - at least here -
now and forever?
Please feel free to suggest a different term.
There is no such thing as a "misdirected bounce".
I thought the term appropriate: we are talking about "bounce" messages,
and they are "directed" to inappropriate mailboxes. "Misdirected" seemed
to fit...
There are only bounces to abused and forged sender addresses.
MTAs sending bounces to those addresses are in full compliance to each
and every RFC.
I quite agree that the MTAs sending these bounces are directing them
exactly as the RFCs specify. (I'm not sure, though, that such activity
can any longer be justified -- I've been on the receiving end of rather
too many recently.)
No MTA that I know of sends bounces to some random address instead of
the envelope sender addresses (which would of course be misdirected).
I'd have to question whether "bounces" to random addresses would
qualify as "directed" at all. Regardless, I don't think they'd still
fit the definition of "bounce".
(YMMV, of course...)
... you can NEVER EVER determine at the SMTP level FOR SURE if the
delivery will work.
Agreed. At best, the receiving SMTP server _might_ be able to verify
that a MailFrom mailbox is appropriate for a message it accepts.
... But those try to force other people to stop sending bounces at all
and thus weaken the reliability of the Internet mail infrastructure.
There might be a middle ground -- where a MailFrom could be marked
as "dubious" and any "bounce" generated to it could be marked with a
"possibly forged MailFrom" warning included in the headers.
(BTW, I've recently been receiving clearly-forged emails designed
to look like bounces but obviously not related to _any_ email sent --
not just unrelated to anything a JLC customer sent.)
--
John Leslie <john(_at_)jlc(_dot_)net>