At XML Europe, W3C XML Query working group member Jonathan Robie gave a report on the family of specifications. This follows up on a flurry of recent specifications updates from the group, and from the XSL working group.

Just before WWW 2002, the W3C XML Query working group announced several updated specifications. A closely related specification, XPath 2.0 (the product of joint work by the XML Query and XSL working groups) was released at the end of April, along with the XSLT 2.0 specification. There has been a lot of recent controversy on the complexity of XQuery and its effect on XPath 2.0 and XSLT 2.0.

Robie explained how XQuery is a synthesis of two prior approaches to XML querying: one very data centric and one very document centric. The XML Query Data Model was developed to underlie the query language and is the same data model behind XPath 2.0 and XSLT 2.0. It views XML documents as an ordered, labeled forest of information nodes with distinct node identity. He compared this view to a grove plan, although some classic grove afficionados might find this a bit of a simplification.

Robie admitted that the dense and formal XQuery specifications of the language, which brings together elements from SQL, OQL, XPath and other sources, are emphatically for implementors rather than users. He went on to describe the syntax using several examples.

There is strong integration with W3C XML Schema (WXS). There are many built-in operators and functions accepting WXS types. Many WXS data types have literal representations in XQuery, and there are type constructor functions for others. There is also strong interaction with the WXS notions of validity as expressed in the post schema validation infoset (PSVI).

Robie also presented a proposal for an XQuery update syntax that apparently has some traction in the working group, though it is as yet completely unofficial, and at a very early stage. It is based on a thesis by Patrick Lehti of Software AG, and further work by Robie and others.

When asked whether he had taken a look at XUpdate, Robie said yes, but that he was looking for an update syntax with strong integration into XQuery, including integration into a static type system. He felt XUpdate is a good solution on its own, but limited and an inadequate basis for the XQuery group.

As an aside, Robie polled the audience of 50 or so as to how many were regular readers of XML-DEV. About 5 hands went up.

See Leigh Dodds' report for more on XML Europe 2002

The recent XQuery, XPath and XSLT working drafts are:

• XSLT 2.0
• XML Query Use Cases
• XPath 2.0
• XQuery
• XQuery Data Model
• XQuery Operators