This cuts right to the chase, making the clearest argument for a less
ambitious proposal that moves XML DTDs to document syntax. I'm considering
rewriting the proposal less as a concrete proposal for syntax (which just
seems to generate endless arguments, and which I'm not an expert at anyway)
and more as an exploration of the implications of (and possibilities for)
using XML document syntax for DTDs.
As an outsider, albeit one who's digging through _The SGML Handbook_, I'm
fairly concerned that a lot of the specs are attempting to do too much. I'm
very glad, for instance, that XML-Linking was broken down into a spec for
Linking and a spec for XPointers. XML-Data seems to me to do far too much in
one place, providing at once an XML syntax for DTDs and a powerful set of data
schemas. While I can't see the battles directly, XML-Data seems to have a lot
of people rather annoyed for a variety of widely different reasons. Still, in
the long run, many projects need its capabilities.
Slowing down seems like one answer to this. Build the foundation and then
build the skyscraper. I'm not totally delighted with the current foundation -
DTD syntax - but I can imagine it being a whole lot worse. Maybe the speed of
the standards process is too slow for developers who want it all _now_, but I
think we'd do well to be less ambitious but improve on the foundations we have
now. Reducing the number of ways to achieve the same result is a good way to
do this, as is making the foundation extensible.
Simon St.Laurent
Dynamic HTML: A Primer / XML: A Primer / Cookies