It was a collection of fixes to XML 1.0 none of which was in itself
sufficiently compelling to justify a new and incompatible version. The
feature I considered the most important, ditching Unicode 2.0 in favor of
an evolving set of characters for element and attribute names (as well as
the values of non-CDATA attributes), was incorporated into the 5th Edition
of XML 1.0.
That's actually problematic all by itself. Parsers have no way of
telling from the data they've been presented which characters are valid.
