RE: Can SOAP be called a XML protocol ? [was: why no doc type dec laration and PIs in SOAP?]

> -----Original Message-----
> From: Andrew Layman [mailto:andrewl@microsoft.com]
> Sent: Friday, September 21, 2001 2:39 PM
> To: xml-dist-app@w3.org
> Subject: RE: Can SOAP be called a XML protocol ? [was: why no doc type
> declaration and PIs in SOAP?]
> 
> 
> I agree with Eric van der Vlist's mail.  Subsetting XML for 
> SOAP starts on a slippery slope to having multiple, incompatible dialects 
> of XML. If a profile of XML is thought needed for certain classes of
device, then
> this should be explicitly addressed as an XML issue, not a SOAP issue.

I agree in principle, but the W3C has thus far shown no interest in
refactoring / subsetting / profiling XML.   

SOAP 1.1 implicitly defines a subset of XML, and that fact has hardly even
been noticed.  I don't think that puts us on a slippery slope, I think it
defines some solid footholds to ensure SOAP interoperability in a rather
muddy environment:

-- Full conformance with XML 1.0 by actual parsers is very rare
-- Details of how the specs mesh with one another are murky and contentious
   (e.g., the DOM and XPath/XSLT data models, or the conceptions of
namespaces in
   the namespace and schema specs, or what non-validating parsers are
supposed to
   do with information in DTDs).
-- Very few people even claim to understand what the "Post Schema Validation
Infoset"
   really is, in what order operations are applied to produce it, and how it
   can serve as the foundation for a new generation of specs (e.g., XQuery).

I'd be happy to have a quid pro quo with the XML Core folks -- you refactor
XML+namespaces+xml:base into full/basic/tiny conformance levels, and we'll
support those conformance levels in SOAP. Until then, I think the status quo
-- with the error handling rules for DTDs and PIs clarified -- is best.

Received on Friday, 21 September 2001 15:27:16 UTC