RE: Use cases

David Cleary is correct. One needs to distinguish between what the DOM API
provides and what the current XPath implementation does (literal strings)
and what the semantics of relative namespaces is in the context of schemas
and XPath/XSLT extensions (indocument retrieval of schemas, extended
functions).

The problem is not necessarily the automatically generated namespaces for
retrieving XML schemas or XSLT extensions by the SQLXML or other MS tools.
Absolutation of them relative to the document they are in is fine from the
schema/XSLT extension processing point of view (disallowing relative
namespaces is certainly not ok). The problem is, that people may chose or
may have chosen to make use of the literal interpretation of namespaceuri
comparisons for their own use over which we do not have control. They
authored their documents according to a valid W3C rec. If we go and change
that rec, the correction should not break their existing documents.

So Simon, I worry more about the hand-authored documents containing inline
schemas in this case, and not the generated ones. Our tools that consume
relative URIs for the above mentioned retrieval purposes do not necessarily
break under absolutation. However, other people may do more with these
relative namespaces than just use them for resolution. Whatever semantics
they add to the relative namespaces besides referring to a schema or some
XSLT extension is outside of our control and may break. Is this unfortunate:
yes, should we break them: absolutely not.

To repeat John's mantra: "Documents are precious; software is not."

More details/corrections to David's misrepresentation of Microsoft's
position below.

Best regards and good night
Michael

> -----Original Message-----
> From: David G. Durand [mailto:david@dynamicDiagrams.com]
> Sent: Wednesday, May 17, 2000 2:07 PM
> To: xml-uri@w3.org
> Subject: RE: Use cases
> 
> 
> At 1:51 PM -0400 5/17/00, David E. Cleary wrote:
> >  > Except that this is not exactly the situation, as I have heard it
> >>  explained by other Microsoft representatives. The _meaning_ of
> >>  relative URLs as  used by Microsoft software is based on  absolutizing
> >>  the relative URL with respect to the document base and using it to
> >>  retrieve a resource. The comparison semantics defined by the
> >>  namespaces spec. are in fact ignored by this software. The namespaces
> >>  specification defines the matching of namespace URIs for identity,
> >>  and does not mandate or endorse any resolution strategy.
> >
> >This is in contradiction to the detailed message a Microsoft
representative
> >posted in the previous discussion within the W3C. Absolution is done only
in
> >those cases of retrieval, not in comparison.

This is correct. See above and the xml-plenary archive.

> This may be so. I was part of the internal W3C discussion, and I must 
> say that Microsoft representatives made a variety of claims and 
> espoused a variety of positions as to the best solutions. I'm happy 
> to hear that I may be wrong, but I think we now must wait for an 
> official description, should one come along. 

For our current official position, see the straw poll vote in the
xml-plenary archive. Since you took part, you should see that all four
Microsoft reps taking part in the straw poll actually agreed upon a single
position. This is rare enough :-).

> None of the discussions 
> that I saw implied that the software in question ever _does_ compare 
> namespaces for identity, as the software did not need that facility. 

This is not the issue. The issue is that people can author documents where
they make use of both features: schema retrieval AND DOM processing where
they perform some literal comparison.

> If comparison is implemented in that way, and actually has an 
> operational effect, then that means that good practice when using 
> that software will include the avoidance of relative URLs as they 
> will have different effects under comparison and retrieval.

Why? The user can well decide that he uses the schema name (which (s)he can
chose in authored docs) to be used for both without leading to any
anomalies.

> Personally, I support making relative URIs for namespaces explicitly 
> deprecated now, and illegal in the next rev.

Illegalization is a very bad thing for users of the existing rev, since it
restricts currently valid behaviour without clear alternatives.

> I don't think the Microsoft, or any other vendor, would actually 
> hesitate a microsecond before changing a data format (and requiring a 
> software upgrade) if it involved potential income rather than the 
> creation of a clean design.

No, I am not going to take this flame bait...

> That said, I believe that the W3C-internal decision to preserve 
> literal comparison, and note the flaws in the use of relative URIs is 
> an adequate compromise. I am actually rather disturbed that we are 
> rehashing an old, acrimonious, and repetitive debate, on which a 
> decision has already been reached.

Here I fully agree.

Received on Wednesday, 17 May 2000 19:48:24 UTC