SOAP message infoset

I had an Action Item from the Feb 13, 2003 telcon to summarize an
interesting thread that cropped up in the XMLP Attachment Requirements
Task Force activity.

One of the proposed requirements for the concrete attachments spec is
R29, currently captured as follows:

  R29. [This requirement engendered a lot of discussion and has
        some significant ramifications for the Abstract Attachment
        Specification and for the basic conception of the SOAP
        message infoset.]

    (a) A message with all its parts, however separated physically, must
        be representable as a single infoset.

    (b) A message with all its parts, however separated physically, must
        be describable as a single XML element in an XML schema.

The following threads give a sense of the discussion that has ensued
on xml-dist-app@w3.org (the public mailing list for XMLP).  I'm
listing only Martin's emails which respond to and quote the other
email on the topic.

 [1] http://lists.w3.org/Archives/Public/xml-dist-app/2003Jan/0051.html
     (The original requirement suggested by Martin Gudgin from Microsoft.)
 [2] http://lists.w3.org/Archives/Public/xml-dist-app/2003Jan/0057.html
     (A reply from Martin to Sanjiva Weerawarana) 
 [3] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0002.html
     (A reply from Martin to Chris Ferris)
 [4] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0003.html
     (Another message from Martin)
 [5] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0004.html
     (A reply from Martin to Rich Salz)
 [6] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0015.html
     (A reply from Martin to John Barton)
 [7] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0025.html
     (Another clarifying comment by Martin)
 [8] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0037.html
     (Another clarifying comment by Martin)
 [9] http://lists.w3.org/Archives/Public/xml-dist-app/2003Feb/0039.html
     (A reply from Martin to Chris Ferris)

There was further discussion on the topic in the Feb 12, 2003 XMLP
telcon with no definitive consensus yet on what to do with the
requirement.  Many people were intrigued by the idea of unifying all
of the information traveling with a SOAP message in a single infoset.
It makes a complete XML Schema/WSDL description potentially simpler
and extends the power of the SOAP processing model to features.  There
are, however, many open questions about how to represent binary
information in this abstract infoset, how signatures would work, how
all of the extra-envelope bits of SOAP 1.2 technology (Webmethod, SOAP
action, etc.) would be normatively specified, etc.

There was some concern that although the idea was appealing, it is too
late to propose this for SOAP 1.2.  Its scope and implications are
obviously much larger than the concrete attachments spec.

Even if the XMLP group does not endorse this model for integration
into its SOAP 1.2 work, it seems like the ideas are worth considering
from an architectural point of view and of possible relevance to
follow-on work in SOAP.

--mark


Mark A. Jones
AT&T Labs -- Strategic Standards Division
Shannon Laboratory
Room 2A02
180 Park Ave.
Florham Park, NJ  07932-0971

email: jones@research.att.com
phone: (973) 360-8326
  fax: (973) 236-6453

Received on Friday, 14 February 2003 15:15:54 UTC