Re: @vocab

On 03/24/2010 01:07 PM, Sebastian Heath wrote:
> I'm responding to Manu's twitter call for comment. I've read many of
> the posts but not all so apologies if any of my points have been
> covered.

Thanks for taking the time to make your thoughts on some of the newer
RDFa 1.1 feature discussions known. We rely on the community to give us
feedback and let us know if the specification is progressing in the
right direction, so your feedback is certainly appreciated. :)

For those of you that aren't familiar with Sebastian's work, he has
written some really interesting posts about publishing Mediterranean
ceramics information using RDFa:

http://mediterraneanceramics.blogspot.com/2010/01/rdfa-patterns-for-ancient-world.html

It was one of those use cases that I hadn't spent much time thinking
about, but is a fantastic example of RDFa in action.

>  I am concerned about how any rdfa tokenizing mechanism will overlap
> with the mechanism for defining new @rel values in HTML5. To be sure,
> this is one of the weirder parts of the standard. Anybody can stick
> anything  http://wiki.whatwg.org/wiki/RelExtensions and it's a
> candidate for being an official @rel value. How are parsers going to
> keep up with that? Let alone distinguish between those values and
> default values. FYI, this is discussed at section 4.12.3.19 of
> http://dev.w3.org/html5/spec/interactive-elements.html . Or am I
> misunderstanding something here?

This mechanism for specifying rel-extensions is heavily debated, but
let's assume that the wiki-based rel-extensions method is accepted as
the standard mechanism for defining new rel values. The moment a value
goes from "proposal" to "accepted", it would be placed in the default
RDFa Profile document for the language of concern (in this case HTML5).

So, for example, since rel="me" has been marked as an "accepted" value
in HTML5, then the HTML5+RDFa Default Profile document is assumed to
have this value in there. Assuming Shane's "rose" proposal[1], the
default profile document for HTML5+RDFa 1.1 would have to be a super-set
of the default profile document for XHTML+RDFa 1.1. I doubt that
XHTML+RDFa 1.1's default profile document would include "me" as a
default keyword... so it may turn out that different languages use
different Default RDFa Profiles.

The key point here is that RDFa Processors will either:

a) Have to dereference the default profile document to figure out the
proper rel values, or have the values hard coded in the processing logic
in some way. We're assuming that the rel values aren't going to change
that greatly from year to year.

OR

b) We'll map any rel/rev value to the Default RDFa Profile for the
language without checking.

Each approach has it's benefits and drawbacks, but there is certainly a
technical solution to this issue. We'll discuss this more when we get
into what/how to specify the Default RDFa Profile document.

> I have also vaguely thought about analogies with xslt extensions.
> Along the lines of the presence of
> 'xmlns:rdfa="http://w3.org/rdfa-processor"' allows one to create
> triples that affect the behavior of the rdfa procesor. Change "rdfa:"
> to whatever you like so long as it's mapped to a specific uri. So
> <link rel="rdfa:object" href="http://vocab.org"/> would mean all
> unprefixed objects would fall into vocab.org . As with all rdfa the
> default @about is the document itself. In the body of document, blocks
> can be wrapped in spans (though that seems very ugly).

This is more or less what @vocab does. We have also defined the "rdfa:"
vocabulary in the latest RDFa spec.

However, even doing what you say wouldn't allow us to do keyword mixing
of ontologies. Not having RDFa Profile documents seem to be a point of
pain for beginning authors. It would also help vocabulary publishers
that mix ontologies to provide vertical-specific vocabularies (selling
digital music, baking recipes, etc.).

Even if this stuff blows up in our face, there's still plenty of time to
correct these recent attempts at making RDFa more powerful and
friendlier for beginners before Last Call (October 2010).

-- manu

[1]http://lists.w3.org/Archives/Public/public-rdfa-wg/2010Mar/0218.html

-- 
Manu Sporny (skype: msporny, twitter: manusporny)
President/CEO - Digital Bazaar, Inc.
blog: PaySwarming Goes Open Source
http://blog.digitalbazaar.com/2010/02/01/bitmunk-payswarming/

Received on Sunday, 28 March 2010 01:09:03 UTC