W3C home > Mailing lists > Public > xml-dist-app@w3.org > March 2002

Re: Need new MEP for SMTP binding

From: Mark Baker <distobj@acm.org>
Date: Thu, 14 Mar 2002 12:25:17 -0500 (EST)
Message-Id: <200203141725.MAA03114@markbaker.ca>
To: skw@hplb.hpl.hp.com (Williams, Stuart)
Cc: xml-dist-app@w3.org
> Hi Mark,
> 
> Not fundementally opposed to your suggestion... but there are a couple of
> buts....
> 
> First question comes back to whether doing an email/smtp/pop binding as a
> proof-of-concept or as a real fully spec'd. For proof-of-concept reuse of a
> feature (the RR MEP) seem's like a good thing.

I agree, and would liked to have reused it (an IMAP binding would have
allowed us to reuse it, FWIW), but being a staunch anti-tunnelist, I'm
not interested in doing request/response over SMTP. 8-)

> I guess that on a
> proof-of-concept basis creating an additional MEP would also exercise the
> framework.

I agree.

> Ironically, I rather like the one-way with correlation/causality captured in
> the AM and would have like that to have been our first 'documented' MEP
> rather than request/response - but we chose to do request/response. Email
> 'Message-ID' and 'In-Reply-To' headers give a means for marking causality.

Exactly right.

> On hop-by-hop acknowledgement, I think that's a QoS difference in that the
> disposition at the sender of sending a message may be: known success; known
> failure; or indeterminate.  More reliable infrastructures provide more
> certainty, less reliable infrastructures gives more indeterminate results.
> It's also interesting that you focus (as would I) on the hop-by-hop nature
> of the acknowledgement (if not the whole MEP) infering ultimately successful
> delivery from hop-by-hop acknowledgements would not be a good idea, they
> represent more of a transfer of responsibility to the next SOAP Node.

Very well said, though I believe the only indeterminate states would be
when;

- the SMTP server never returns a response
- the connection to the SMTP server drops after the message is sent, but
before the response is received

So I'm not sure why "more reliable infrastructures" would be able to
yield less indeterminate results, because their frequency is primarily
a function of the network, not the software.  No?

MB
-- 
Mark Baker, Chief Science Officer, Planetfred, Inc.
Ottawa, Ontario, CANADA.      mbaker@planetfred.com
http://www.markbaker.ca   http://www.planetfred.com
Received on Thursday, 14 March 2002 12:34:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:59:09 GMT