Bernie Cosell <[EMAIL PROTECTED]> writes:
> Just to make clear here, in the course of lauding AOL for not bombing
> third-parties, keep in mind that this is just a consequence of
> *their*own* non-compliant email handling choice, and is _not_ inherent
> in "good email handling". Granted, they handle a flood of email that is
> unimaginable by almost anyone, still: they [in spirit, if not in letter]
> violate the protocol [by SMTP-accepting a message they end up having no
> intention of keeping] and then run into the problem that there's then no
> good way to after-the-fact reject/bounce the message.
There's nothing in the slightest bit protocol-violating about that. Lots
of mail systems do that for a wide variety of reasons; gatewaying and
firewall systems have been doing things that way for years. If you have a
huge internal network, the firewall machine accepting mail from the
outside world may not have any idea whether a given address is valid.
Stanford's mail system has been doing that since we introduced
@stanford.edu aliasing in order to be able to provide a fuzzy match bounce
(which in practice is greatly appreciated by our users). qmail does this
all the time due to its security model; the SMTP listener doesn't know
what addresses are valid, since obtaining that information requires
privileges that the network-exposed portion of the program doesn't have.
--
Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/>
