Re: My opinion on the problem of identifying ID semantics

I think there are two steps to be considered in any possible TAG 
endorsement of xml:id.

1. Should W3C take steps to create a normative specification for xml:id?

2. If so, what guidelines should be given for use of xml:id in future W3C 
recommendations, as well as in other (I.e. non-W3C) vocabularies?

The purpose of this note is not to offer an opinion on #1, but rather to 
suggest that #2 needs some attention.  I've been thinking about SOAP's use 
of its own id-like attributes.  I note that most proposals for xml:id 
effectively have document-scope, which is right for many use cases.  The 
SOAP attributes are not intended to have such scope.  They are for use in 
constructing graphs that can encompass some of the content in a document, 
but not the rest.  There might be other reasons which someone might want 
to put identifiers on elements, and those might indeed have document 
scope.  Thus, it is incoherent to put a soap-enc:id on the soap:envelope 
element.  Question 2 asks:  were we to be starting on the specification 
for SOAP encoding today, would advocates of xml:id recommend that we use 
it in place of soap-enc:id? 

I think that answering #2 is useful in its own right, and is also 
revealing of the underlying merits of doing xml:id at all.  Again, it is 
not the purpose of this note to offer an opinion on that one way or the 
other.  Thank you.

Noah

[1] http://www.w3.org/TR/2003/PR-soap12-part2-20030507/#uniqueids

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

Received on Thursday, 22 May 2003 13:50:48 UTC