Re: AD review comments on timely delivery

> At 07:32 PM 9/15/2002 -0400, Keith Moore wrote:
> >I stongly object to trying to redefine email delivery delivery in terms
> >of retrieval from the mailbox.  Delivery is when the message becomes
> >accessible from the mailbox.

> There are two different issues.  One is concept, the other is terminology.

> Noting when a message makes it to the recipient, rather than merely being
> deposited into the recipient's mailbox, is the key conceptual point.  It is
> an essential distinction, because there is a significantly different sense
> about the recipient's "involvement" in the message.

> In non-computing life, we refer to both situations as "delivery", but we
> use different qualifying language to make clear where (or to whom) the
> message was delivered.

> Note that POP is a DELIVERY protocol, not an access protocol.  From a
> formal standpoint in email architecture, picking a message up with POP
> effects final delivery.

I agree that's how POP3 is often used, but it most certainly isn't how it is
defined or implemented. Like it or not, POP3 is defined and sometimes used as a
mailbox access protocol, with delivery occurring before that point.

RFC 1939 is entirely concerned with "maildrop access". It says nothing about
POP3 being involved in final delivery.

An additional problem is that in most implementations delivery to the store is
independent of subsequent access to the store. This modularity is in some ways
a benefit, but when it comes to the delivery agent imposing additional
semantics on activities beyond the point of final delivery it is anything
but.

				Ned