Re: on independence of elements, relating versions [XMLVersioning-41]

Hi Dan,

On Wed, Mar 03, 2004 at 04:30:25AM -0500, Dan Connolly wrote:
> At the W3C Tech Plenary, Dave Orchard raised discussion
> of
> 
> "When is "must ignore" reasonable for non-rendering semantics?"
>  -- http://www.w3.org/2004/Talks/0303-tagext/slide5-0.html
> 
> in extensibility and versioning, and Henry Thompson
> made a point that ignoring only works when the semantics
> of elements are independent from each other.
> 
> Another speaker mentioned the importance of explicitly
> describing relationships between versions.

Interesting.  The latter better matches my mental model, but I'd bet
that both are roughly equivalent in some important way.

> It occurred to me that in a way, RDF/XML syntax
> takes the principle of independence of elements to an
> extreme. And the whole purpose RDF schema mechanisms
> like rdfs:subClassOf and rdfs:subPropertyOf is for
> describing relationships between versions.

Absolutely!

If RDF had mandatory extensions it would be able to describe more of
these kinds of relationships.  I've done some investigation into this,
and pretty much concluded that while it doesn't seem too much work to
add (though I don't know if it messes up the model theory at all), the
*really* hard part is going to be getting support into RDF tools.  So
I'm not optimistic we can have this any time soon, though at least from
a protocol POV it's a pretty simple roll-out; just mint a new media type
per serialization.

> I think the TAG should advise XML vocabulary designers
> that if this sort of versioning is important to them,
> they should give RDF a serious look.

Strongly agreed.

> I hope to draft something soon. Has anybody out there
> already got something relevant? Anybody want to help?

I've written bits and pieces on this subject for the past 3+ years,
primarily in weblog and email form.  I did gave a presentation[1] a
couple of months ago which describes the value of RDF in a novel
(AFAIK) manner; as providing XML a self-descriptive extensibility model.
Perhaps that sort of approach to describing the value to folks might be
fruitful; I had one convert at the talk at least. 8-)

I'd be happy to help in any way I can.

 [1] http://www.markbaker.ca/Talks/2004-xmlself/slide1-0.html

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca

Received on Wednesday, 3 March 2004 08:52:50 UTC