Re: A new approach to accomplishing RDFa Profiles

Hey Mark,

good you provoked me to try to collect my thoughts:

On 2010-3-20 16:28 , Mark Birbeck wrote:
[skip]
> 
> I realise that some people don't like the fact that prefixes and
> tokens now occupy the same space, but I do feel that we're now at a
> point where the onus should be on those people to come up with more
> than just 'I don't like it', or 'authors won't get it'.
> 

and you are right that the issue is not really technical. I therefore
tried to formulate my reasons to myself, too; here it goes...

Yes, it is correct that there is an entirely consistent model that can
be argued for. What it says is that the xmlns syntax (or its
equivalents) define, in effect, an alias for a uri. That uri could be
used by itself in an RDFa attribute; and we can reformulate the CURIE
description which says that a:b means that one takes the uri for which
'a' is an alias, and concatenates it with 'b' (let us put the bnode
CURIEs aside for now). So yes, this _is_ consistent. And, technically,
it is indeed not a radical departure from what we have.

However. For good or for worse, we chose to use the @xmlns syntax in the
past. And even if I know the usage is not the same as XML Namespaces,
for most of the people the @xmlns _is_ a namespace declaration where, as
far as I know, it is not possible to do with the alias what you
describe. We have developed a specification, primers, tutorials, etc,
around the idea of the prefix notion through @xmlns for RDFa1.0.

What this means is that, if we adopted that new model, I feel it would
mean we would have to look at all the past recommendations and tutorials
very closely to make the necessary changes in presenting the uri alias
model. If we do not do that, this would create confusion among RDFa1.0
adopters. Although it is true that if somebody comes to the RDFa1.1
world only, for that person the alias model would be entirely clear and
consistent. The confusion would be with our old customers.

I could then turn around and ask you: what would we gain by doing so?
The current separation of prefixes and keywords is no less consistent,
we know it is possible to define what we want in, say, @profile files,
it is entirely consistent with all our past documents in terms of
presentation, so why bother changing the model? I can see no reason for
doing so, but we would buy ourselves extra work...

Cheers

Ivan


-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF   : http://www.ivan-herman.net/foaf.rdf
vCard  : http://www.ivan-herman.net/HermanIvan.vcf