By the start of May we'll have published a number of important speficiations; we can expect last call working drafts for some XSLT and XML Query documents, advancement of XPointer and XInclude, progress on XML 1.1 and namespaces 1.1, and I'll talk about the impact of all these.

Updating XML itself with XML 1.1 (blueberry) has been a long and difficult process, and still isn't finished. Implementation experience has been promising, however. XPointer got lost in the wilderness but has emerged leaner (really!) and fitter. XInclude was waiting for XPointer, and provides a useful alternative to declaring external parsed entities in many environments.

The XML Core Working Group has been requested to examine several new issues recently, including XML-instance syntaxes for declaring text entities, handling of external non-parsed entities without using notation declarations, identifying ID-valued attributes, and even DTD-less profiles or subsets of XML.

All these point to a general move away from a lexical view of XML processing and put the infoset, DOM, or data model first. I'll talk a bit about this, and about how our roots in document processing are still important to us.

As always with XML, everything can be represented as a stream of characters that any other processor can understand. Keeping that principle and yet satisfying the needs of the various groups asking for change is a diffcult and challenging balance, but ultimately a very rewarding one.