Re: IDs? and classes

At 07:15 PM 6/13/2000 -0500, Al Gilman wrote:
>The bad news is that, based on the discussion that just went on about URIs
>in XML on the xml-uri@w3.org mailing list, I would estimate the chance that
>the XML implementers would buy the dodge that Charles suggested (putting a
>URI-reference to the definition of "what I mean by this class" as the value
>of the class attribute) as being roughly comparable to the chances of a
>snowball surviving in hell.
>
>The good news is that they are beginning to come arond to the idea that
>they need to come up with some plan that fills this functional gap if they
>aren't going to have W3C cram something they don't like down their throats.
>
>What we have in HTML is the html.profile parameter.  This is by definition
>a reference to data description for the usage in the META element.
>
>I continue to claim that if the resource linked to via the html.profile
>parameter contained a data dictionary for CLASS values as well, it wouldn't
>break anything and would be the maximally HTML-backward-compatible way to
>introduce additional knowledge keyed to CLASS marks.  This is [my current
>working estimate of the best way] to introduce user-defined markup in the
>HTML context.

This gives a solution to HTML, but we need to use this in other languages 
as well. It would be nice to have a consistent solution.


>As far as IDs are concerned, they are good from a legacy perspective
>because they are already in XML 1.0.  On the other hand, XML processors in
>future will comprehend XPath and we will be able to generate third-party
>information for documents where the IDs are not provided on the items or
>ranges we wish to annotate.

Right, by using XPointer we can point to any place in a document even 
without IDs. However, having ID's at least in the main elements will help 
in resolving the right place to point to even when the document has been 
changed as changes happen often without our knowledge when we annotate 
somebody elses documents in the Web.

Marja

>  The match-patterns for more-info rules can be
>a mix of stuff the author put in the document and generic  relationships
>from the DTD or schema that they were using.
>
>Al
>
>At 06:01 PM 2000-06-13 -0400, Wendy A Chisholm wrote:
> >Charles,
> >
> >Are you suggesting something along the lines of <P
> >class="http://foo.bar/definition.htm"> ?
> >
> >That does not seem to be the most elegant way to do things.  Isn't there
> >some way to use RDF?  Namespaces?  something else?  I agree that the URI is
> >helpful for the person who want to find out about the semantics, but how
> >would this be machine-understandable?
> >
> >I like Marja's original idea of include ID's on elements.  ID's could be
> >arbitrary and automatically generated for block elements.  Then,
> >annotations could be attached to any element in the document.
> >
> >--wendy
> >
> >At 04:41 AM 6/12/00 , Charles McCathieNevile wrote:
> >>Actually, in the context of the "semantic web", and RDF, I have a 
> suggestion
> >>to make, which is that classes be used which are URIs - prefereably real
> >>ones. This would enable two things to happen:
> >>
> >>  1. An author could explain, at the URI in a dereferenceable document, 
> what
> >>the class was about or for.
> >>
> >>  2. It would become more or less trivial to make RDF assertions about
> >>classes, and therefore about how to re-use existing ones rather than create
> >>new ones for each piece of content.
> >>
> >>In general, I am opposed to making a class if it can be avoided (for
>example,
> >>it is better to use the existing CODE element than to produce a style class
> >>for delineating code examples). In particular I would suggest that the
> >>semantics of map were not extended in HTML 4.01, merely the syntax, which
>was
> >>extended to match in the real world the semantics of the specification. But
> >>that is a trivial question I guess.
> >>
> >>cheers
> >>
> >>Charles McCN
> >>
> >>On Mon, 12 Jun 2000, Jason White wrote:
> >>
> >>   Interestingly, there has been significant resistance, within this 
> working
> >>   group, to any attempt to provide common semantics to specific values of
> >>   the HTML CLASS attribute, either within the guidelines or techniques
> >>   documents. The basic rationale was that the semantics of CLASS values
>were
> >>   left completely unconstrained by the HTML specification and it was
> >>   desirable not to create an inconsistency, or apparent inconsistency,
> >>   between HTML 4.0 and the guidelines. It was also urged that content
> >>   developers should have total freedom in creating style sheets to use the
> >>   CLASS attribute as they wished.
> >>
> >>[and so on]
> >
> >--
> >wendy a chisholm
> >world wide web consortium
> >web accessibility initiative
> >madison, wi usa
> >tel: +1 608 663 6346
> >/--
> >

Received on Thursday, 15 June 2000 17:13:59 UTC