What Is the Product of this Task? (WAS RE: [Minutes] 2 Dec 2002 TAG teleconference (New issues: XML subsetting, Binary XML, metad ata in URIs))

That XML-SW has received little comment or become 
controversial in the XML-Dev list might be taken 
as a sign that you are right.  Still, I am a little 
unclear as to what this task (whoever does it) is:

1.  When you say, "subset XML", is everyone clear 
that this means a) "XML as defined in the XML 1.0 
specification", or does this mean, b) "subsets of 
the XML family of specifications including XML 1.0 
plus....".  Without being crystal clear here, this 
effort is likely to take a very long time.

2.  Is the effort otherwise an effort to create 
a means to indicate when subsets of a and/or b are 
to be supported by an implementation of an XML 
application? 

3.  Are these means intended to be used in the 
application instance or in the application specification?

It is clear by existence proof (the SOAP specification) 
that some developers need to use less of XML 1.0 and 
that they indicate this in their specification.  It 
is clear that an XML 1.0 compliant processor would be 
unaware of the application constraints.   It is clear 
that this would affect interoperability should a 
naive developer expect an XML 1.0 compliant processor 
to be aware.

If the specification is to include:

1.  Subsets (proper subsets of XML 1.0)

2.  Profiles (cross-products of the proper subsets 
of XML 1.0)

this seems to be straightforward.   If these 
subsets and profiles are to include other specifications 
(eg, Namespaces, base), then the job is more complex 
and there will be a natural tendancy to want to 
include the task of creating a new version of 
XML 1.0 (eg, XML-SW) because this would be 
the sensible way to redefine the meaning of 
"XML compliant processor".   If the only task is 
to indicate in an application specification what 
parts of various XML specifications including 
XML 1.0, the task is simpler but not much more 
useful than the current specifications.

So my intuition here is that some want to redefine 
what an XML compliant processor must support.  Is 
everyone really ready to do that and take on the 
costs of the implementation and fielding of that? 
That will be controversial.

len


From: Paul Grosso [mailto:pgrosso@arbortext.com]

I am not yet convinced that there is widespread
agreement that it makes sense to subset XML.

paul

Received on Wednesday, 4 December 2002 15:43:07 UTC