Re: RDFa Lite and non-RDFa @rel values

On Apr 25, 2012, at 16:25 , Alex Milowski wrote:

> On Wed, Apr 25, 2012 at 2:31 AM, Ivan Herman <ivan@w3.org> wrote:
>> Wow. Lots of mails while I was on the road yesterday...
>> 
>> This is a set of replies to a number of issues in one place; collecting those may help moving forward.
>> 
>> ---
>> 
>> Alex, you said:
>> 
>> [[[
>> That means, the real question comes down to what we expect from this
>> kind of markup:
>> 
>>  <a vocab="..." href="http://www.w3.org/" rel="nofollow"
>> property="homepage">W3C's Home Page</a>
>> 
>> Should it be:
>> 
>>  <> <http://www.w3.org/1999/xhtml/vocab#nofollow> <http://www.w3.org/>
>>  <> <...homepage> "W3C's Home Page"
>> 
>> or
>> 
>>  <> <http://www.w3.org/1999/xhtml/vocab#nofollow> <http://www.w3.org/>
>>  <> <...homepage> <http://www.w3.org/>
>> 
>> ?
>> ]]]
>> 
>> Well, in my view, none of the two. I believe the response should be:
>> 
>> <> <...homepage> <http://www.w3.org/> .
>> 
>> And nothing else.
> 
> In HTML5 with the HTML or XHTML syntax?  

Both HTML5 and XHTML5.

> That's where I start to get
> heartburn with all of this.
> 
> I've been detailing solely with XHTML documents and so my perspective
> is skewed.  The initial context is different and so link relations all
> come out nicely as RDFa triples.  I've been enjoying the fact that
> they do as link analysis via RDFa is very useful.
> 

Indeed, XHTML1 keeps the old set. This is one of the main (probably THE main) difference between HTML5+RDFa and XHTML1.1 + RDFa

> The fact is now, with both of these documents, that if you take
> exactly the same document and serialize it in the XHTML and HTML
> syntaxes, you'll get different RDFa triples.  That bothers me even
> more than this current issue.  You can fake some of it by setting the
> @vocab attribute but that won't help once you've used it for your own
> vocabulary.  You just can't prime the "local default vocabulary"
> yourself--only the host language can.  As such, we've treated XHTML
> and HTML syntaxes as different host languages when they spring from
> the same specification and that seems to be going the wrong direction
> as HTML5 has attempted to bring them closer together.

I think the sentiment in the group, and definitely the feedbacks we got from different users, was that the automatic usage of the XHTML1 @rel value was not very popular (a typical case of pollute the space with triples originating from rel="stylesheet"). (At least that is my impression.) The only reason we kept the XHTML1 version was backward compatibility...

> 
>> 
>> The issue here is that, as we looked at this before, the predefined @rel values make very little semantic sense for RDF. That is the very reason that we did _not_ generate any triples automatically for those guys in HTML5. Hence the proposal of Stéphane to, essentially, drop those from the generated RDF is actually in line with this observation, because they are _not_ meant for RDF. (The user always has the possibility to use a CURIE or a full URI if he/she _wants_ a triple to appear there.)
> 
> I think that's a matter of perspective.  Just because a @rel value
> doesn't help marking up small structures (e.g. a person) doesn't mean
> they don't have value overall for the relationships between documents
> on the web.
> 
>> ---
>> Then there is DanBri's comment that I would like to take very seriously. We need finalization and stability NOW; we are getting to "The perfect is the enemy of the good" effect here. Any delay may seriously backfire as for the acceptance of RDFa. I know this is not a technical argument, but that is where we are...
>> 
> 
> I understand the seriousness of the schedule.
> 
>> Bottom line: I am still in favour of Stéphane's option 1. The only other alternative I can live with is to stay with the status quo and move on.
> 
> Putting my W3C hat on, doesn't "fixing" that require testing for
> non-expansion in section "7.5 Sequence" result in a substantive change
> that requires going back to CR?
> 

With my W3C hat on:-) the extra processing on removing @rel values is defined for HTML5 only. It does not change the core processing steps. HTML5+RDFa language profile is defined as a recommendation by the HTML5 WG, and that is still further down the line. So we do not have a process issue for this.

Cheers

Ivan

> 
> 
> -- 
> --Alex Milowski
> "The excellence of grammar as a guide is proportional to the paucity of the
> inflexions, i.e. to the degree of analysis effected by the language
> considered."
> 
> Bertrand Russell in a footnote of Principles of Mathematics
> 


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