According to Dave Sill:
> Well, ignoring the fact that this "fundamental concept" isn't codified
> in the relevant RFC's, how do you think things should work? Should
> there be some univeral "max SMTP sessions limit between any two
> hosts"? What should it be? One? Ten? However many a typical sendmail
> system can handle? Wouldn't it be a whole lot easier if the receiving
> system just refused connections when its plate got full? (Doesn't
> sendmail already do this when the load average gets too high? Maybe
> sendmail needs a "MaxSMTPSessions" option).
It just doesn't make any sense to split apart multiple recipients
before the final destination. Consider how antisocial this is
if you are delivering to a large list of users whose machine is a
uucp hop away from their smtp gateway. There is a reason that
multiple recipients are allowed per message. Why would anyone
intentionally defeat it?