On Dec 1, 2:17am, Eric S. Raymond wrote:
}
} Bart Schaefer <schaefer(_at_)brasslantern(_dot_)com>:
} > What I disagree with is the claim that the *sending* system is in violation
} > of RFC2821 because the return-path has become invalid. Put another way, I
} > don't agree that RFC2821 *justifies* your refusal of the message, merely
} > that it does not *prohibit* such refusal.
}
} You are correct. The path may *become* invalid without putting the
} sending MTA retroactively in violation. It must however *be* valid
} at time of transmission. That is the plain meaning of 4.1.1.4.
You mean 4.1.1.2? "The reverse-path consists of the sender mailbox."
} We are simply maintaining that sending MTAs MUST issue a valid return
} path at time of transmission.
And I assert that even that is a new requirement. Valid from what point of
view?
Suppose the (not uncommon case) of a sending MTA that is not the same host
as the MX for the sender's domain. Call these MTA1 and MX1. A message is
transmitted via MTA1 to the recipient's MX, which I'll call MX2. If the
sender mailbox is valid at MTA1 -- that is, MTA1 can successfully deliver
a DSN to that mailbox in the event of a 5xx response from MX2 -- where is
the requirement that the mailbox also be valid when presented as a RCPT TO:
at MX1?
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg