I don't know if the level idea is useful or not, but what I had in mind is that
once "Minimal HTML" ( = level0 HTML) is defined, implementors can build things
like HTML->text translators (for mail gateways?), linemode browsers, and minimal
functionality tools like that. Then folks can validate (using the Minimal HTML DTD)
that their docs will work with such tools.
Similarly, we could take what Mosaic 2.2 (or 2.4?) implements, find a suitable
DTD (close to the 1.8 version I sent out), and call that level1. And so on...
If there are various "markets" for HTML documents, then we need various HTML
levels. On the other hand, if all the authors need the same features, and all
the implementations support them equally well, then there's no need for various
"levels."
[This raises the question: what do level0 browsers do with level1 docs?
I hate the "ignore tags you don't recognize" idea. But it may be the
only way to go...]
>The Maths extensions are a lot of work to implement and the only browser to
>implement them implements very little else. But they are absolutely essential
>if we are ever going to publish papers on the web as anlternative to
>paying sums to get journals to publish papers then buying back the finished
>product, doing all the work en-route for free making huge profits for
>various publishing houses (eg that of Bob `anyone for a swim' Maxwell).
>
>So I can see that putting maths into another level would be a good idea.
>
Do we expect to be able to do interactive typesetting of mathematics? Pretty
ambitious. If I were tackling the problem of integrating math into WWW, I'd
be working on the "multiple MIME types in one data stream" problem, and use
LaTeX some such for creating/editing math, and then develop tools to typeset
LaTeX into encapsulated postscript, CGM, Windows MetaFile, Mac PICT format
and the like (along with support for bitmap formats, since X doesn't support
any MacDraw style graphics formats)
>
>Before that happens I think that people will have to go to full SGML or
>at least a configuration file which is what mine uses. SGML fans might like
>to note that a DTD is useless for doing this :-(
A DTD is useless for doing what? Expressing various levels and availability
of various features? Check out
http://www.hal.com/%7Econnolly/html-test/html.dtd
$Id: html.dtd,v 1.8 1994/04/09 01:02:10 connolly Exp $
for a counterexample. Right now, I can validate that a document uses
* only "standard" features (read: features I like)
* only Mosaic-supported features
(credit for the idea of using "feature test entities"
goes to Dave Raggett, who got it from the TEI folks)
Dan