Re: The Web Services Architecture WG position on XML profiling/subset ting

I respectfully request that this note be withdrawn, on process grounds.

This is appears to be a formal statement of position by the WS-Arch group 
to the TAG as to where XML, SOAP and XML Schema should go.  It contains a 
number of statements about SOAP itself that I believe to be factually in 
error.  It also proposes moves such as normative schema subsets, which is 
a topic that has received much careful discussion over a 3 year period 
within the Schema workgroup.

My concern is that, as a member of both the Schema and the XMLP workgroups 
I don't believe that there was any attempt by WS Arch to work with either 
of the WGs to check their facts or to see whether pertinent investigations 
had been done that should be considered.  I find this particularly 
troubling given that WS Arch and XMLP are in the same Activity and share a 
CG.  I hope the W3C still emphases consensus as a cornerstone of its 
process, and consensus can only be achieved by communication and 
coordination.

Whether my concerns about the facts or conclusions are "correct:" is not 
the main point.  I believe that when a workgroup formally approaches the 
Tag with a "conclusion" it should be done with some care, and with a real 
effort to first coordinate with other pertinent workgroups in the W3C.  At 
the very least their should be a "heads up" alert in advance, giving the 
workgroups a chance to say:  "we have something to say about this, we 
don't have consensus on your interpretation of our spec, etc."  I think 
this particularly true in the case of closely related workgroups such as 
XMLP and WSA.

I specifically do not propose to debate the technical concerns here, 
because my point is that this isn't the place.  Some of the areas where I 
disagree will be clear to anyone who reads my recent contributions to the 
tag list.  Factual concerns includes include statements such as that SOAP 
"punts" on defining IDs in the absence of DTDs.  Not so.  A normative 
definition is at [1].  Note in particular the constraint at [2].  The fact 
that it's done at the SOAP level rather than using some more general 
purpose mechanism is a design decision one could question, but to call it 
"punting" doesn't seem to be a constructive addition to the discussion. 
Please just reference the normative prose and state your concerns.

I suggest that the Tag in coordination with the appropriate workgroup 
chairs and CGs clarify the appropriate process to be used in this case. If 
I am off base in raising these concerns, please accept my apologies.

Thank you.

Noah

P.S. We need to decide where to continue this discussion, and I don't have 
any great ideas.  I specifically propose that technical discussion between 
WSA, XMLP and Schema move off the tag list until such time as we decide 
how to proceed.  Might I suggest that we all stay quiet for a few minutes 
on the technical questions until our chairs, the WS CG and/or the Tag can 
signal whether my concern is deemed appropriate from a process point of 
view, and if so how they'd like to proceed?  I suppose that process 
responses should at least be on the Tag list unless the Tag directs 
otherwise, but I don't know how to avoid cross-posting and still reach 
those likely to be concerned.

[1] http://www.w3.org/TR/2002/CR-soap12-part2-20021219/#idattr
[2] 
http://www.w3.org/TR/2002/CR-soap12-part2-20021219/#uniqueidconstraints

------------------------------------------------------------------
Noah Mendelsohn                              Voice: 1-617-693-4036
IBM Corporation                                Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------







"Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>
Sent by: www-tag-request@w3.org
01/17/03 03:35 AM

 
        To:     www-tag@w3.org
        cc:     www-ws-arch@w3.org, (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        The Web Services Architecture WG position on XML profiling/subset ting
Categories: 
 






The Web Services architecture WG wishes to makes its position known to the
TAG on issue xmlProfiles-29 " When, whither and how to profile W3C
specifications in the XML Family".  As you know, this issue came to your
attention because SOAP (since its inception) has forbidden DTD internal
subsets, external DTD references, and processing instructions.  The 
reasons
for this are very well stated in
http://lists.w3.org/Archives/Public/www-tag/2002Dec/0119.html and.  We do
wish to state this working group's opinion that it is vital for the W3C to
take action to formally recognize the profile of XML that is commonly used
in Web services applications so as to promote the use of generic XML tools
in web services applications. We strongly believe that the "muddle 
through"
option being advocated by some on the www-tag list will set into concrete 
a
situation that is sub-optimal at best for the Web services community and
bound to foster a disconnect between the Web services and XML communities.
 
One solution, which the SOAP community has pioneered, is to "bless" a
profile of XML technologies as the foundation on which to build.  In the
case of SOAP, that includes a subset of XML 1.0 that allows composability 
of
XML messages,  plus the Namespaces spec, and the XML Base spec.   It would
be very desirable to see this profile given authoritative status by the 
W3C
so that XML tool developers could either develop parsers, etc. that are
optimized for this profile or allow the user to check conformance to the
profile.   This alleviates a problem today in which a SOAP message can be
valid according to the SOAP 1.2 schema but be illegal with respect to the
SOAP specification itself, e.g., if it contains a DTD internal subset.  
The
current situation hinders interoperability, the reuse of generic XML tools
for Web service development, and sows confusion among the users.

One issue that SOAP 1.2 "punts" on (as do some other W3C specifications,
including DOM Level 2 and XSLT) is the question of how an "id" attribute 
is
defined in the absence of a DTD and without insisting on a schema 
processing
step.  Some officially sanctioned mechanism -- ID attributes in the XML
namespace ("xml:id"), an ID attribute declaration mechanism suppported in
the core of XML (e.g. "xml:idattr="ID"), or even an approved convention 
that
"ID", "id" and "Id" are all to be considered ID attributes unless the ID
attribute is specified by some other mechanism, seem like viable options 
at
this point.

Another issue of concern to the Web services community is xml:base.  SOAP
uses URIs for some identifiers including, but not limited to, values of 
the
encodingStyle (see 5.1.1 SOAP encodingStyle Attribute) and role (see 5.2.2
SOAP role Attribute) attribute information items. SOAP does not define a
base URI but relies on the mechanisms defined in the XML Base 
specification
and [RFC 2396] for establishing a base URI against which relative URIs can
be made absolute.  Thus, it is important for an XML processor used to
process SOAP messages to implement the xml:base specification, and this
should be reflected in the WS-friendly XML profile.

Additional features that would make XML more widely useful for Web 
services
have been advocated by some members of the Web Services Architecture WG. 
We
note that they are not supported by the same overwhelming majority of the
the WG endorsing this message.  For example, it would be very desirable to
formalize a profile of the XML schema specification that is suitable for 
web
services applications; it would be desireable to further enhance the
composability of XML documents, e.g. by a standardized mechanism allowing 
an
XML parser to handle files containing multiple well-formed XML fragments 
but
without an enclosing element, and to allow multiple character encodings to
be used within a single XML document.  We realize that there are practical
challenges preventing these features to be agreed upon quickly, and they 
go
beyond the sentiment to NOT add any additional features in the profile
discussion, and suggest merely that these be considered at some point in 
the
future.

There is currently no strong opinion within the WSA WG on how the profiles
should be specified, whether in a future rev of the XML spec or an adjunct
specification. The important thing to us is not the specific mechanism or
even the details of a profile, but encouraging a widely-adoped convention
that will allow XML tools to guarantee conformance with the profile of XML
that has come into widespread use in the Web services community.

Received on Friday, 17 January 2003 09:54:12 UTC