W3C home > Mailing lists > Public > xml-dist-app@w3.org > February 2001

Re: Thoughts about path and intermediaries

From: Martin Gudgin <marting@develop.com>
Date: Fri, 9 Feb 2001 17:33:25 -0000
Message-ID: <007d01c092be$69a9ef80$0200a8c0@greyarea>
To: "Mark Nottingham" <mnot@akamai.com>
Cc: <xml-dist-app@w3.org>
Sorry... off the top of my head...

A path is an ordered list of nodes through which an XML Protocol message
passes on it's way from sender to ultimate recipient.

The entire path or parts of it could be implicit as in the case of an
HTTP/SMTP gateway that is part of a given system configuration. Any XML
Protocol messages sent to an SMTP destination from an HTTP based sender
would pass through that gateway. The sending application doesn't need to
know about the gateway.

The entire path or parts of it could be explicit as in the case of a sender
specifying that an XML Protocol message must pass through a node that
provides a digital message signing facility ( to use what seems to have
become our de facto example :-) ). In this case the sending application is
aware of the path ( or parts of the path ) that the XML Protocol message
will take. The XML Protocol also needs to be aware of the path so that it
can make sure the message actually gets sent via the correct node.

So I guess there are several factors we could build a matrix out of...

1.    Is the path explicit, implicit or a mixture of both?

2.    Is the XML Protocol layer aware of the explicit path ( if any )?

3.    Is the XML Protocol layer aware of the implicit path ( if any )?

4.    Is the application layer aware of the explicit path ( if any )?

5.    Is the application layer aware of the implicit path ( if any )?

FWIW - I ( currently :-) ) think the answers are ( or should be ); both, no,
yes, yes, no. Which I think pushes all the intermediaries which process
parts of the original message generated by the sending application up into
the application layer. Which makes XML Protocol single-hop with respect to
processing with multi-hop being an application level construct. Note that it
still allows for an XML Protocol message to pass through multiple nodes ( to
switch transports for example ) just that no processing of the message
occurs at those nodes.


----- Original Message -----
From: "Mark Nottingham" <mnot@akamai.com>
To: "Martin Gudgin" <marting@develop.com>
Cc: <xml-dist-app@w3.org>
Sent: Friday, February 09, 2001 4:09 PM
Subject: Re: Thoughts about path and intermediaries

> Sorry, I think I missed the definition of 'path' (which seems pretty
> key here). Is there a reference?
> On Fri, Feb 09, 2001 at 12:47:51PM -0000, Martin Gudgin wrote:
> > Having spent some time thinking about this Stuart and I have come to the
> > following conclusion;
> >
> > If the XML Protocol Layer directly supports the notion of a path then we
> > support processing intermediaries that sit between the sender and the
> > ultimate recipient. We can also support the targeting of XML Protocol
> > Modules at particular XML Protocol Handlers located at those processing
> > intermediaries.
> >
> > Conversely if the XML Protocol Layer does NOT support the notion of a
> > then it becomes inherently single-hop. In this latter case path becomes
> > application level construct and not part of the core definition of the
> > Protocol. This would simplify the core definition of XML Protocol while
> > still allowing applications to layer intermediary processing on top of
> > Protocol.
> >
> > Thoughts, comments, flames etc. to the usual address
> >
> > Gudge and Stuart
> >
> >
> --
> Mark Nottingham, Research Scientist
> Akamai Technologies (San Mateo, CA)
Received on Friday, 9 February 2001 12:34:29 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:01:11 UTC