I really think we're going the wrong way here. HTML was always meant
to be a small, even minimal document type. The promised strong point
of the web would be multiple document types, all tied together within
a structure of HTML documents.
HTML only provides the framework by which lots of other document types
are accessed. You want to do lot's of neat TeX things? Use TeX. You
want to do something with integrated text and graphics? Use Postscript.
You want to define a single document as a collection of different
things? MIME does this for you. On the other hand, if we make it a
de facto requirement that everything available thru the web must be
HTML, I think we limiting ourselves severely (and being somewhat
MIME is supposed to be where all of our work is going right now. By
developing a good MIME document viewer, and a good transport mechanism,
we can do all this (and come up with interface to read MIME mail and news
while we're at it). And we set up a system which can be easily extended
to document types that haven't even been designed yet.
Last, let me make another pitch for HTPS. HyperText PostScript. HTML
provides the structure, and other document types provide the data. But
sometimes it'd be nice to have the structure and the data all together.
It would be trivial to add an operator to postscript which defines anchors
as arbitrary paths associated with URLs. That should satisfy most anyone's
desire for arbitrarily composed documents. If anyone is interested in
implementing this, contact me, I have some good ideas.