Re: Why not XHTML+RDF? was Re: Links are links

----- Original Message -----
From: <DPawson@rnib.org.uk>
To: <timbl@w3.org>; <DPawson@rnib.org.uk>; <jonathan@openhealth.org>;
<paul@prescod.net>; <Norman.Walsh@Sun.COM>
Cc: <www-tag@w3.org>
Sent: Tuesday, October 08, 2002 10:12 AM
Subject: RE: Why not XHTML+RDF? was Re: Links are links


> TBL wrote:
>
> > There is a boundary over which RDF could be used.
>
>   I'd prefer a wall, a high one. But then that's being facetious. Sorry.

Apology accepted

>   RDF is for machines IMO, not manual editing.

As is any XML :) . But I agree that RDF/XML is

> > A solution for example is to make a hypertext link
> > using html:a  or xlink:href or whatever to indicate the
> > hypertext nature of the link, but then to use RDF to express
> > the relationships between various alternative long texts in different
> > languages,
> > which is the sort of thing RDF is natural for.
> >
> > This would move longdesc out of HTML's purview into
> > a very reusable metadata vocabulary which could be used in
> > all kinds of different ways with different specs.
> >
> >     <html:img src="car.png"/>
> >     <rdf:Description rdf:about="car.png">
> >         <wai:longdesc rdf:resource="car_en.txt"/>
> >         <wai:longdesc rdf:resource="car_fr.txt"/>
> >         <wai:audioalt rdf:resource="car_en.wav"/>
> >     </rdf:Description>
> >
> >     <rdf:Description rdf:about="car_en.txt"
> >         wai:summary="A black convertible a little the worse for wear">
> >         <i18n:lang>en</i18n:lang>
> >     </rdf:Description>
> >
> >     <rdf:Description rdf:about="car_fr.txt">
> >         <i18n:lang>fr</i18n:lang>
> >     </rdf:Description>
> >
> > With longdesc moved out, one of the problems of using xlink
> > (or hlink)  goes
> > away:
> > that of trying to do attributes on attributes and many links from one
> > element. And the problem of adderssing more  complex
> > multilingual/multmedia
> > cases
> > is solved.  And it kinda blends with cc/pp coming form the
> > other direction.
> >
> > So this would leave a link itself as a simpler thing.
>
>
> And make the chances of people using wai ns'd stuff about nil?
> Thanks Tim.

Well, it is interesting to think about whether a general quite powerful WAI
and general multimedia solution would be less or more deployed
than LONGDESC which is by many accounts rather a failure in terms
of deployment.

If one is mentally in a one-namespace world, then one tries to
get everything one needs into it, everything one doesn't need
out, and live within it.  If one is are mentally in a multi-namespace
world, then you want nice clean modularity between namespaces.

Having different solutions for HTML and SVG will make life
more difficult when documents mix them.

> Just look at the crap in the way in your example.

I'd point out that the example does what cannot at all be attempted
with HTML, so you have nothing to compare it with.

If you are using XML, then things are by design verbose.
A non-xml syntax would remove a certian amount of verbosity,
it is true.

     <car.png>     :longdesc <car_en.txt>,  <car_fr.txt> .
     <car_en.txt>
        :summary "A black convertible a little the worse for wear.";
          i18n:lang "en" .
     <car_fr.txt>     i18n:lang "fr" .

> It'd scare the pants off 98% of web authors.
>   Nice ns based solution, but if we're asking for manually generated
> content, which is the case here for longdesc, then thanks, but no thanks.

Well, you may be right, but you may also be wrong.
It's difficult to judge.  People pick up the weirdest things just by copying
other pages.
It's not a question of what to use for longdesc -its a question of
how to extend it for more complex things.

> When are you going to pull the xml namespace (family silver) out?
>   Surely a hypertext link is about as fundemental as it could get?

Sometimes your efforts to be sarcastic (and, incidentally, sarcasm
is *not* something we need on this list) sometimes cloud your meaning.
I'm afraid couldn't follow what you were suggesting there.


> Regards DaveP.


IMHO
Tim BL

Received on Friday, 11 October 2002 19:24:12 UTC