At 10:17 -0400 on 06/28/2005, Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu wrote about Re:
Bounce/System Notification Address Verification:
On Mon, 27 Jun 2005 21:10:38 EDT, Hector Santos said:
Keith, we were referring (atleast I was) to the fact that they do EXIST 1
RCPT limits for a NULL return path. There are plenty of systems that work
in this mode whether you care to believe that or not. The conflict you seem
to forget is that your multi rcpt-to/null return path allowance is a source
of spam.
No, what *I* said was that since, *in general*, a bounce only has 1 MAIL FROM
to generate RCPT TO's, and there's a lot of idiot mail admins that feel
justified in saying "Since this is true 98% of the time, we will force it
to be true always...."
I will rephrase that to be "Since in 98% of the time there is only
one Recipient in a <> message, if the sender has some need/desire to
send a multi-recipient <> message, then such sender should create a
separate copy for each recipient and thus send them as
single-recipient copies". Thus puts the burden of respecting the
one-recipient-per-<>-message "rule" where it belongs - on those
processes that generate those 2% of the <> Messages by requiring them
to make the effort to conform (ie: By sending multiple
single-recipient copies).