I've been down that path (ala XML-DATA) but we have people writing DTD's
who are skilled at document analysis but not overly comfortable with
destructuring the content model of an element and converting it to some
form of XML instance. It increases the chance of error, and generates
extra QA/fix cycles. Our analysis has shown that error minimization
mechanisms are the most significant contributor to the sustainable
profitability of software (or software-like) companies at our level.
I actually would rather this approach myself, and use it whenever I
access the DTD within msxml. The skill level wouldn't be an issue if I
had time to write a good DTD authoring tool. Nothing I have seen comes
close to cutting the mustard.
+----------------------------------+
| Antony Blakey |
| N-Space Pty Ltd |
| Java - CORBA - SGML - XML |
| mailto:antony@n-space.com.au |
| http://www.n-space.com.au |
+----------------------------------+