W3C home > Mailing lists > Public > public-rdfa-wg@w3.org > March 2010

Re: attempting to merge the 'vocab' and 'profile' documents

From: Ivan Herman <ivan@w3.org>
Date: Tue, 09 Mar 2010 10:01:10 +0100
Message-ID: <4B960E56.2050608@w3.org>
To: Ben Adida <ben@adida.net>
CC: Mark Birbeck <mark.birbeck@webbackplane.com>, W3C RDFa WG <public-rdfa-wg@w3.org>
Ben,

If we forget about the 'how we do it' for a moment, I think having a
mechanism to put the zillions of xmlns: statement into one place and
replacing it with one reference is important. In my personal view it is
even _more_ important than the keyword mechanism.

Indeed: let us not think in terms of small vocabularies here;
vocabularies can be large. As an example, we are referring here to
Dublin Core fairly often; but let us not forget that in their latest
incarnation, ie,

http://purl.org/dc/terms/

there are much more terms than before: 55 (it used to be 15 in their
'legacy' space). Would we require the vocabulary publishers to publish
separate RDFa vocabulary files to publish separate keywords URI-s, too,
beyond the RDF files they already publish? This simply does not scale
for vocabularies that may hundred or more terms...

Coming back to 'how we do it'. Yes, the current proposal merges the two.
And though that is technically correct in the sense that it works, I was
wondering myself whether we should not keep the two separated. I almost
had a second version of the document doing that but then I decided not
to do it before having a discussion. What I had in mind is:

- in the vocabulary files there is a separate RDF term for prefixes and
for keywords

- the modification on the CURIE processing is different: the current
document proposes a modification whereby column and resourceless CURIE-s
simply map to the the URI that is mapped by that prefix; if we separate
the two than this would be restricted to terms that have been explicitly
identified as keywords

I have a slight preference, actually, for that technically more complex
approach because it might be cleaner for the users. Besides, if we want
to restrict the usage of keywords to certain attributes (see my mail on
that[1]) than such a separation might actually become necessary. But all
this is open for discussion.

Ivan

[1] http://lists.w3.org/Archives/Public/public-rdfa-wg/2010Mar/0034.html


On 2010-3-8 23:35 , Ben Adida wrote:
> On 3/7/10 1:52 AM, Ivan Herman wrote:
>>
>> Hm, this is not what I feel is happening. The processor is working on
>> some non-RDF data (XHTML+RDFa file) and, the 'vocab' graph directs this
>> processor on how to generate that 'target' draft.
> 
> I *think* I see what you're saying Ivan, and if so I think I lean in
> your direction.
> 
> Allow me to raise one issue that I believe is related and that I don't
> think has been raised yet. Given my non-attendance on calls, I'm going
> to try to do this rarely, so I'm raising this now because I think it may
> be quite important.
> 
> *** Should we really allow the @vocab/@profile document to define
> *prefixes*, rather than just keywords? ***
> 
> In other words, no doubt we want this:
> 
>   <div profile="http://astrology.org/vocab#">
>     <span property="sign">Pisces</span>
>   </div>
> 
> But do we really want this:
> 
>   <div profile="http://astrology.org/vocab#">
>     <span property="astro:sign">Pisces</span>
>   </div>
> 
> which feels odd because now we're saying that RDFa 1.1 markup is going
> to regularly throw CURIE resolution errors with an RDFa 1.0 parser.
> 
> I understand that, from Mark's point of view and implementation
> proposal, there's no difference between defining a prefix and defining a
> keyword, but I'm not sure that's natural to most people: it requires
> buying into the idea that property="foo" means a "foo" *prefix* and no
> suffix, rather than an empty prefix and a "foo" suffix, which is the
> *much* more natural way to interpret how xmlns is typically handled. And
> it now means that @xmlns in RDFa, rather than being a simple
> augmentation of @xmlns in RDF/XML, is now actually quite different.
> 
> All this to say: I know we *can* build a solution where prefixes are
> defined elsewhere... but do we want to? Do we really need to? I think if
> we say "use xmlns if you want to use prefixes, use vocab if you want to
> use bundles of keywords", we've got ourselves 90+% of the use cases, and
> a lot less complexity.
> 
> OK, flame me :)
> 
> -Ben

-- 

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



Received on Tuesday, 9 March 2010 09:00:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 04:55:06 GMT