> Well, I guess what surprises me is that there's only one error
> message rather than two. For whatever reason, nsgmls is totally
> ignoring zero-length CDATA sections rather than following the rules
> from clause 3.2.1.
It doesn't surprise me that hammers cannot turn screws very well.
Remember that NSGMLS, like the PSGML editor, is a general-purpose SGML
tool that happens to be able to handle XML documents without gagging.
XML imposes additional constraints beyond SGML, some of which are
stated explicitly and some of which are implicit in the (very
considerable) differences between the ISO 8879 and XML 1.0 syntactic
productions.
Since neither NSGMLS nor PSGML actually uses the XML 1.0 syntactic
productions for parsing, both miss some XML-specific errors (and are
thus non-conformant), though they shouldn't report spurious errors.
If you're interested in purity-checking, you need to use a tool that
is actually built on top of the XML 1.0 productions, like James's
Expat.
All the best,
David
--
David Megginson david@megginson.com
http://www.megginson.com/