Paul Burchard (
Wed, 7 Jun 95 05:54:44 EDT

Brian Behlendorf <> writes:
> On Mon, 5 Jun 1995, Paul Burchard wrote:
> >While I like <LINK REL=Script>, it's too global to be the general
> >solution for scripting. Think about VRML, and it'll be especially
> >clear that you want be able to take advantage of the hierarchical
> >document structure and bind the script to a subdocument.
> Absolutely... but I think you'd want to be able to do it
> without explicitly marking that element you want to
> modify/animate in the text itself, which is why a call for
> a new tag doesn't ring true to me. There are several ways of
> marking up HTML so it could be subaddressed using the ID
> and CLASS attributes on most tags in HTML 3. If there's a
> standard way of referring to "All paragraphs in CLASS
> 'foo'" or "the <FIG> with ID 'bar'" in HTML 3, then
> couldn't that be used by the script to determine both
> input data and output areas?

The CLASS/stylesheet approach is a great idea! I had been
picturing mostly IDs as the way a script would find specific
elements of its subdocument to act on.

However, I think you'll still want some kind of hierarchical
scoping. This is necessary to enable the proliferation of small
useful scripts. If all scripts have global scope in a document,
just imagine the oppressive melee of conflicts that will result!
This hierarchical scoping capability is, in essence, the purpose of
the proposed <COMPOSITE> (<C>?) linking element.

> I get the feeling that plugging turing machines and
> document languages together isn't going to be easy

Well, I guess we do know how to perform that binding for linearly
structured documents of 1's and 0's, at least... :-)

Paul Burchard <>
``I'm still learning how to count backwards from infinity...''