However, this is such a stupendously big change that there's no point
thinking about it until HTML 4, I think, or later.
WHy do I say that?
You have to add the following:
* Entity management
* Ability to declare entities in the document header
* somewhere for the HyTime to point to:
[1] nameloc, using an ID -- ths is the same as <A NAME=...>; no point
adding only this.
[2] treeloc -- requires that browsers construct a document tree instead
of a flat linked list; requires (Worse!) that all browsers construct
the _same_ tree. In other words, if one browser implies a <P> where
no tag was given, and another doesn't, a HyTime treeeloc that counts
elements will go wrong in one or other browser.
[3] dataloc -- requires that the browsers all deal with whitespace in
#PCDATA context in the same way. Including RS/RE. See the Exoterica
white paper for some clearly written notes about RS/RE (i.e. CR and LF).
[4] HyQ -- out of the question :-)
[5] notloc -- a URL can be written as notloc, and a trivial DTD change
already posted here (last year) makes the existing HTML method of
using URLs be valid HyTime, without requiring that the documents change.
Of course, this does not give new functionality.
I really don't think the web is ready for this.
I admit that SoftQuad Panorama has some support for HyTime; however, it
deals only with valid, parsing SGML documents. Possibly Panorama could
be used to gain experience in this area, though, by working with (valid)
HTML documents in Panorama. If anyone is interested in pursuing this,
let me know and some time in the next few weeks I'll make some software
available for you.
But let's not hold up an HTML 2.1 or an HTML 3 for HyTime.
Lee