TODO LIST:

	* Handling of external entities (both param and normal) in Prolog
	* Check ID/IDREF
	* Donot use quoted values for tag identifications.  Right now the
	  following leads to bad error handling:

		<tag name="value>
	* Allow for (a,b) as attribute-type (name-group)
	* Handle source-info in included SYSTEM entities

UNICODE:

	* Started full unicode support under the CVS branch XML_UNICODE.
	* Old version only allows for unicide in CDATA (attribute values
	  and element content)
	* New adds throughout: element-names, tags, filenames, etc.
	
	ISSUES:
		- Port to Windows
		- Testing
			- File-entities
		- Leak setting filename using istrdup() in sgml2pl
		- Verify performance.  Optimise.
		- Own character classification and conversion
			- Copy from Prolog?