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

Re: mid-course correction on abstract model for module processing

From: Mark Jones <jones@research.att.com>
Date: Wed, 4 Apr 2001 10:26:00 -0400 (EDT)
Message-Id: <200104041426.KAA06749@glad.research.att.com>
To: hugo@w3.org, xml-dist-app@w3.org
	Date: Wed, 4 Apr 2001 09:10:43 -0400
	From: Hugo Haas <hugo@w3.org>
	To: xml-dist-app@w3.org
	Subject: Re: mid-course correction on abstract model for module processing

	* Jean-Jacques Moreau <moreau@crf.canon.fr> [2001-04-04 14:21+0200]
	> R802 may indeed lead to some representation that physically
	> partitions the message, and header/body is one such representation;
	> but then it only helps the final receiver "process XMLP blocks
	> intended for them without processing the entire message", *not* the
	> intermediaries.

	I might be missing something here; both intermediaries and the final
	destination can have:
	- a well-defined set of blocks addressed to them and arranged so that
	  they do not process (i.e. parse) what's not theirs. For
	  intermediaries, they can stop once they know that the rest is for
	  the final destination.
	- references to other blocks that may be out of this set (as pointed
	  out by Mark) that defeat this ordering.
	- references to trailers that would be require them to parse the whole

	Could you explain what difference between the final receiver and an
	intermediary makes you say that?

	I am unsure about what is intended by "processing" in R802... Since
	Henrik talked about it for the Header/Body distinction, I am assuming
	that it really means "parsing".

	Hugo Haas - W3C
	mailto:hugo@w3.org - http://www.w3.org/People/Hugo/ - tel:+1-617-452-2092

Here's what I think is the point of the header/body issue and R802.  A
simple and perhaps common scenario would have the Body containing
blocks that represent the bulk of the bytes being exchanged between
sender and ultimate recipient.  In this case, the hope is that
intermediaries might be able to do a detailed parse of only a small
portion of the envelope (the blocks in the Header), identifying and
processing those header blocks targeted at the current node, and just
copying through the rest of the message (from <Body> on).

I'm inclined to agree with Hugo that R802 probably meant "parsing"
rather than "processing" since intermediaries should only be
processing blocks that are tageted at them anyway.  The wording in
R802 is sloppy in this regard.  Furthermore, the efficiency that R802
implies isn't ensured by the SOAP Header/Body design since header
blocks can reference body blocks (thus requiring that the Body be

Even if R802 means "parsing" rather than "processing", it is not clear
that the header/body syntax is required for efficiency.  There may be
other means for skipping (and copying through) blocks that are not
intended for the intermediary without full parsing.


Mark Jones
AT&T Labs
Received on Wednesday, 4 April 2001 10:26:02 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:01:12 UTC