[Asrg] is SMTP still really store-and-forward ?

[Steve Atkins]

On Nov 28, 2008, at 1:03 PM, David Nicol wrote:

> On Fri, Nov 28, 2008 at 2:17 PM, Chris Lewis <clewis at nortel.com>  
> wrote:
>> Your list article traversed _four_ store-and-forward hops after  
>> leaving
>> the IETF mail server before getting to me - and those are just our
>> internal MTAs.  And it looks like it went thru several store-and- 
>> forward
>> hops getting from your account on gmail to the IETF, and a couple  
>> inside
>> the IETF.
>
> I said "realm" not "host." From the list to all of our in-boxes, the
> envelope return address was politely changed to
> "asrg-bounces at irtf.org" by Mailman version 2.1.9.
>
> Figure 1 in rfc821 is the SMTP model, and it does not involve
> intermediate hops.  Complex multi-host systems one one side or the
> other collapse into "Sender SMTP" and "Receiver SMTP" in that diagram.
>
> gmail realm <-> irtf realm <-> nortel realm
>
> Two transmissions between realms.
>
> I meant "realm" not "host" which is what I said.  If there's a better
> term for what I mean by "realm" I am open to changing my terminology.
>
> In arguing to drop store-and-forward as a requirement, the internal of
> operations of a realm are considered a black box outside the FUSSP's
> scope.

If there is store-and-forward at any point along the path, that'll break
most toy ideas that are broken by store-and-forward.

If your toy idea is not like those other toy ideas then you should  
really
talk about the details, rather than asserting that the store-and-forward
that's commonly in use is not really store-and-forward.

Or, at least, define terms better. As one example of that, are you  
asserting
that the recipient MX always knows everything there is to know about
whether an email will be successfully delivered or not, such that all
the store and forward hops beyond that can be safely ignored?

Cheers,
   Steve