Re: RDDL proposal from Sandro Hawke

Sandro Hawke wrote:


>
> > Sandro Hawke wrote:
> > > Let's just use the existing "rel" and "type" attributes on a-href
> > > elements.
> >
> > This is elegantly simple and minimal, but has one shortcoming: there's
> > no obvious way to determine which href= is that of the purpose and which
> > is that of the nature.  I.e. I want to fetch a RDDL and say "find me a
> > related resource which is an XML Schema and is appropriate for run-time
> > validation"  What's the solution to that?  Do we have to have just one
> > reserved attribute? -Tim
>

The idea of using <a rel="..." type="..."> to replace <rddl:resource
xlink:arcrole="..." xlink:role="..."> has the advantage in that it doesn't
'clutter' an XHTML document with non-XHTML namespace elements.

In the earliest days of RDDL (http://www.rddl.org/#background) we'd
considered overloading the <a> element, but decided that introducing a new
<rddl:resource> element would be better as it would be clearer that it was a
new element with a new intended use, and would not confuse folks who use
HTML that it was some type of hypertext link i.e. we felt that html:A was
not intended as a generic link, rather Xlink would be a better option.

Shrug. That's the way we had seen it. I can also see some advantages to not
needing a new DTD/schema etc. for RDDL, not needing to deal with XHTML
modularization etc. On the other hand, I'd rather fix XHTML modularization
and keep RDDL as an example of multinamespace'd documents hopefully done
right.

I'm also not terribly keen on overloading the QName in attribute value idiom
with _another_ way of binding prefixes to URIs (arrghh). Other than that,
the proposal *is* elegent and minimal.

Jonathan

Received on Wednesday, 22 January 2003 21:51:23 UTC