> Surely the unanswerable argument against CDATA elements in XML was
> they prevent you from parsing a document without the DTD. Just like
> optional start/end tags, and unmarked empty elements.
A good reason, but you could always say that "every CDATA element must have
an attribute xml:content-mode='CDATA'". So not unanswerable (though neither
<![CDATA[ ]]> nor the attribute commend themselves). And not unthinkable, as
xml:lang and xml:space prove.
Rick Jelliffe