RE: [xsl] keys and idrefs - XSLT2 request?

["Michael Kay" <mhkay at iclway dot co dot uk>]

> But without wishing to engage you in debate, could you just clarify a
> couple of fundamentals?
>
> > Firstly, XPath 2.0 is being developed jointly by the XSL WG and the
> > XQuery WG.
>
> I was under the impression from Jim Melton's mail of 9th September
> (http://www.biglist.com/lists/xsl-list/archives/200109/msg00537.html)
> that XQuery 1.0 and XPath 2.0 were two different languages, albeit
> based on the same data model and functionality.
>
> I was therefore under the impression that the wonderful XSL WG would
> be in charge of the development of XPath 2.0 while the slightly
> suspect XQuery WG would be in charge of the development of XQuery 1.0.
> It gave me hope that XPath 2.0 could be kept simple and elegant for
> the XSLT audience, while XQuery 1.0 could be built with the power that
> would satisfy their audience.
>
> Is it rather that XQuery 1.0 will be a superset of XPath 2.0?

Firstly, if you read the two charters, you'll see that both groups are
tasked to collaborate on XPath 2.0, but the responsibility for publishing
the spec rests with XSL WG:
http://www.w3.org/Style/2000/xsl-charter.html
http://www.w3.org/XML/Group/2001/04/xmlquery-charter

The Query charter states that XPath 2.0 will contain the functionality that
is common to both XSLT and XQuery. You can read that as saying that XQuery
1.0 will be a superset of XPath 2.0; though there will probably be things in
XPath 1.0 that XQuery 2.0 doesn't want (like some of the implicit type
conversions) which leaves scope for making the relationship not quite so
simple.

The collaboration hasn't always been a smooth one, as the two groups have
different backgrounds and each has a lot of diversity within the group. But
I'd say that of the dozen or so difficult technical issues that we found
hard to resolve, we've probably reached consensus on about eight, so I'm
feeling reasonably optimistic that we'll deliver a language that meets both
the functionality and the compatibility requirements. Whether it ends up
being elegant is something we'd better leave for users to judge...

And I'd better say nothing about your caricatures of the two groups. Let the
specs speak for themselves.
>
> > It's certainly likely to be true that if you add a schema to an
> > existing document, then existing stylesheets will in some cases
> > produce different results. For example, if you describe two
> > attributes as dates, then they will be compared as dates rather than
> > as strings.
>
> I had the impression from the XQuery/XPath data model WD that DTD
> information would be treated in a similar way to schemas (e.g. that
> you would be able to tell which attributes were IDREFS attributes and
> therefore break them up into a sequence of IDREF values). Is this not
> the case?
>
I think there is a lot of detail still to be worked out on this. There's a
general assumption that a DTD can act as a sort of minimal schema. But there
are still a lot of options on the table regarding the use of schema
information, e.g. whether it has to be available statically or dynamically,
whether you can choose to ignore the schema/DTD even if one exists, and so
on. Your note on these subjects is a useful contribution: there have been
quite a few other thoughts, but it's all still in the melting pot.

Mike Kay


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list