Draft Note on DOM Support for XPath 2.0

In March the DOM WG produced a candidate recommendation draft of DOM
support for XPath 1.0.  Candidate recommendation means we encourage
implementers to experimentally implement it, and it is less likely to
change than previous drafts.

http://www.w3.org/TR/DOM-Level-3-XPath/

(For the record, we now have one well-known ECMAScript implementation
in all Mozilla browsers and one well-known Java implementation that we
know we can test, and as we have heard of a number of other
implementations we are looking for a second either accessible through
ECMAScript or through Java to meet the requirements so the specification
can become an official recommendation, so please let us know of your
efforts in this area, even if they are incomplete, which we might be
able to test to satisfy implementation requirements).

For a variety of reasons, this draft did not support XPath 2.0.
The 2.0 model had sufficient differences to require different support,
and the XPath 2.0 model was not close to being a recommendation at that
time.  Additionally, the DOM working group was only tasked with the 1.0
interface and has not been chartered to carry out the additional work
required.

But we all realize that XPath 2.0 is important to many and some
implementations would like to support both (we know this is true) or
perhaps even just XPath 2.0.

We produced the attached note on how to extend the Candidate
Recommendation specification to support XPath 2.0.  This keeps all
interfaces but one -- the result object, which is substituted to
support the enhanced types, sequences, etc.  This is why the
specification returns an object instead of a specific result object
type.  The 2.0 request codes are distinct permitting an application to
support either or both and an implementation to support either or both.

The DOM WG intends to eventually submit this as a formal note to W3C,
but we know that people are experimenting today with implementation of
XPath 2.0 on DOM and would like to give a unified starting point of
experimentation and allow you to give feedback.  Notes may be the starting
points for new specifications if there is enough interest by W3C members in
following up in another W3C activity, but whether or not that happens, we
feel that it is good to get this note out there where implementers can see
it.  We may revise it a bit based upon feedback, although this is not
currently on recommendation track.

Let us know what you think, and tell us your experiences and what parts
of the DOM XPath module you have implemented so we can test them.

Thanks,

Ray Whitmer
ray@xmission.com