Hilarie,
I think I finally got the idea:
In this scenario we are really thinking about a proxy like functionality of the
MTA.
A sender MTA does a dialog with the proxy and the proxy is doing a dialog with
another MTA at the same time. Forwarding the RCPT command to the next MTA and
sending back the response.
And here response modification comes into play.
Correct so far?
If the next MTA is returning an error, it could of course be "modified", if the
proxy selects another address and tries that one again in a RCPT command to the
next proxy. On success it sends back the positive response. Is this the way the
"alternative mapping" method would work?
The callout server is determining the other address, maybe even other next MTA.
So it tells the proxy MTA to send another command to the same or a different
next MTA?
Also correct?
Thank you for helping me to understand.
Regards
Martin
-----Ursprüngliche Nachricht-----
Von: owner-ietf-openproxy(_at_)mail(_dot_)imc(_dot_)org im Auftrag von
The Purple Streak, Hilarie Orman
Gesendet: So 23.01.2005 02:01
An: ietf-openproxy(_at_)imc(_dot_)org
Cc:
Betreff: RE: Response modification
On Sat, 22 Jan 2005 at 21:09:22 +0100 Martin Stecher contended:
> > Further, we rule out a large number of useful
> > identity-based services by eliminating the ability to do
> > response modification.
> Here I disagree.
> Tonny describes again a use case that we've listed already last year.
> See sample 2-2 on http://www.martin-stecher.de/opes/smtpusecases.html
> No doubt that it is a very important one.
> This and other examples of group 2 are the "useful identity-based
services"
> you are referring to, aren't they?
> And these are all request satisfaction examples, not response
modification.
> Again: Response modification is only useful if the callout service
makes
> use of the original response data, which in SMTP is the SMTP reply.
> The only service I can think of, that does this, is a logging
service.
> Is that one strong enough to have response modification in the SMTP
deployment?
> Or can you please describe another important service that makes use
of
> the response and modifies it?
When the response comes back indicating that there's a problem with
the recipient name, the OPES processor sends that to the callout
server, and at that time, the callout server can pick an alternative
method for finding the recipient. This seems obvious to me ...
I'm not sure why it seems obscure to others. I wrote a white paper
for Novell several years ago about "global identities" that used this
method for resolving delivery addresses.
Hilarie