TODO list:
        * Integrate widestring support with the PyExpat module (major thing)
	* Switch to 4DOM's DOM implementation 
	* Add SAX2 support
	* Drop wstrop & Unicode; Python 1.6 will handle this
	* Speed up the builder class somehow, and do some performance tests
	* Change HTMLBuilder to use SAX instead of SGMLlib
	* Make more passes over the documentation
	* More demo programs 
	* Add READMEs to existing demo programs
	* Lots of docstrings!
        * Re-indent everything to 4-space indents
	* saxlib.AttributeList should really support all dictionary behaviour
	* Flesh out the test suite more
	* Update the Windows DLLs and integrate Christian Tismer's WISE 
	  installer

demo/xbel/	
	* Update all the software to match the current XBEL DTD
	* Add a test suite for the basic parser

dom/
	* Convert all the raise statement to use the exception(arg) form

	* Implement reading of SGML documents for FileReader

	* DocumentType class is mostly unfinished; what should the
interface for creating them look like?

	* Element.removeAttributeNode needs to know about DTDs to
restore a default value.

	* Walker: merge walk() and walk1() into one function (or at
least make it more generic)

	* XmlLineariser: what should it do with PIs and other similar things?

	* NodeList returned from .getElementsByTagName should be live.
(Hard, and doesn't seem to be very useful; AMK doesn't really care.)

	* DOM Level 2 changes

	* Greg Stein owes doc on qp_xml
