The introduction of multiple namespaces in XML based documents brings to light a number of sets of presentation and interaction naming sets. A single namespace generates a well bounded set of event, scripting and presentation semantics that is well understood. Use of multiple namespaces however, provides a new set of problems in dealing with a number of issues.
The W3C DOM is a well defined model for representing an XML document in memory. In the case where a number of different namespaces are mixed in a single XML document, a number of issues arise in relation to the effects of event propagation, styling across namespace boundaries, conflicting display requirements and so forth.
Architecturally it would be ideal to allow multiple vendors to provide implementation for specific namespaces and in so doing provide a framework in which such development could occur. In order to allow rapid development of multiple namespace documents, the framework should allow indpendent vendors to interact with a notional 'virtual' DOM.
I have some experience implementing an experimental system that uses a plug-in style architecture to let each plug-in handle an individual namespace. These experiments have shown that it is possible, but also that the in-memory representation of the DOM requires each namespace plug-in to own it's in-memory model. Thus the framework itself manages a 'virtual' DOM which is semantically identical to the standard W3C DOM.
My position is that this an area requiring fundamental design control, most likely from a working group. Also, it is a wide-ranging area that should be under the auspices of a group that interacts with all W3C groups such as the TAG.