On Fri, 6 Sep 1996, Daniel LaLiberte wrote:
> Rendering for humans is not the main purpose but it is a significant
> side benefit. The generality of the data structure is a much more
> important goal. The META element only permits embedded values,
> whereas the LINK element only permits references to external data, and
> neither permits more complex grouping or nesting of structures.
> Furthermore, their use seems tailored to this arbitrary collection of
> "meta" data things with no clear definition of how to draw the
> boundary (and no clear boundary possible, in my opinion).
OK, so if we're talking about having a more general structure than META
in HTML 2.0 provides for, what was wrong with the DCES SGML DTD? Why do
we need this in HTML at all if its _not_ specifically for rendering?
HTML is after basically a rendering oriented SGML DTD. What we're after
here is structuring so why not just use a clean DTD without having to
support all the bogosity of <CENTER> and <BLINK>? And no need to worry
about stepping on toes in W3C, etc. We don't need most of HTML for this
and we're having to bend bits of it to make it do what we want.
> Like you, I am assuming it *might* be external to the document. If
> we want to embed this HTML data structure in an HTML document and
> hide it from casual view, then we probably need some other extension
> to HTML. Simply putting it in the HEAD might work, but I think not for
> all browsers. But I think there is much more value in permitting the
> metadata to be external to the document; it can be associated with the
> document via a LINK tag (you suggest META, but how?).
If we embed metadata in HTML documents in META elements where the HTML
DTD > 3.2 (Wilbur), then we can push for META to have additional
attributes to those that it current had and maybe even new elements that
are valid in the HEAD.
> I think it would be a mistake to put a META section in an HTML
> document intended as metadata for *another* document. How would you
> specify the metadata of *that* document, such as who created it and
> when? (Please don't go anti-recursive on me Stu. This is a basic,
> relatively simple concept that most people won't have to deal with
> anyway.)
Good point.
I think we've got several things all going on at once here, and it might
be worthwhile listing them to make sure that we're all talking about the
same things. As I see it we've got (ignoring Warwick Framework for a
moment):
1) Embedding the Dublin Core Element Set into HTML 2.0 using the META
element. I think we reached consensus on this a few weeks back and its a
done deal. It certainly seems OK to me (see the metadata in
<URL:http://www.roads.lut.ac.uk/> for example), and I know its being
written up as we speak,
2) Using a new SGML DTD to hold the Dublin Core Element Set. This is
what Lou, Liam, Eric et al came up with. This can be linked to from
other resources (such as HTML documents using LINK) and is pretty
easy to parse as SGML goes. I think that this is also pretty much a
done deal; I certainly haven't seen much in the way of complaints on the
list about it and its been widely advertised (referenced from a Dlib
article for example),
3) Using HTML to hold some structured metadata (possibly including
Dublin Core) using <DL>..</DL>. This I still think is a bit odd. If
its just to be rendered, why does it need new elements/attributes?
If its for machine parsing of DCES elements why not just use the
HTML 2.0 style META embedding as in 1) or a LINKed document conforming
to the DTD in 2)?
So am I off the crease here? What's missing from/wrong with 1) and 2) that
means we need 3)?
Tatty bye,
Jim'll
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
* I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
|