Charles Chou writes:
> If XML is going to be the language to encode and represent databases, the
> the issue on Query Languages will ultimately be: what model is the target
> database built by. Relational database model and its standard query
> language, SQL became a standard  because it is the only model where data in
> our complex world can be represented in such a way that can be
> mathematically analyzed and understood.

This is a big problem: not only is SQL aimed at the traditional
record-and-field relational database, which is not the text markup
model, its syntax sucks as far as the end user is concerned, which
makes it very inappropriate as a publicly-usable query language for
SGML data.

Obviously there's nothing to stop it being used inside a shell which
shields the user from the lower levels of interaction, such as a query
form or script, but as Charles points out...

> Searching a randomly structured

Nice oxymoron. Bit like my students who come looking for a
carefully selected random sample for a survey :-)

> XML database is a hopeless task, no matter what Query Language one uses.
> The notion of searching for a Web Browser that will "understand" XML and
> display correctly all of the information encoded is just as hopeless.

This is the difficult thing to get across to the user. SGML markup just
provides a bunch of hooks on which you can hang applications. The hype
surrounding XML has made it almost impossible to persade people that it
doesn't do everything.

An interesting sideline is that we now have people doing stuff with markup
and parsers that has been bread-and-butter SGML for a decade, but hailing
it as the new salvation because it's got a new label on :-)