Re: Determination of subjects/objects

Ben Adida wrote:
>> [ a rdf:Seq;
>>  rdf:_1 <#A>;
>>  rdf:_2 <#B>.
>> ]
>> <#A> p:q "bla2".
>> <#B> p:q "bla2".
>>
>> Note that, as a side effect of the Ben rules, there is _no_ difference
>> between @about and @href in this setting...
> 
> Okay, *this* is a problem, I think. I know the way I worded "the rules",
> it appears that @about is then the RDF identity, so it appears that it
> can be the object when you write:
> 
>    <li rel="rdf:_1" about="#A">
> 
> but I think that is very wrong. Because, if you add @resource, then
> @about suddenly becomes the subject. And that's quite confusing.
> 
> It leads me to wonder if the rules are a bit wrong or inconsistent.
> 
> If you were to write resource="#A", then I would agree, but as it
> stands, it bothers me. And, in fact, this is the Achille's heel of this
>  "syntactic sugar for collections" issue: if you add other RDFa
> attributes, you screw up the resulting output.
> 

Ben, I remember I was surprised when realizing that, too. But, just to
really separate the issues, I do not believe this has anything to do
with the collection issue. The very same happens with

<span rel="a:b" about="#b">asasfas</span>

You rules in

http://www.w3.org/mid/46A8D3ED.2080404@adida.net

say:

- - @rel (conversely @rev) creates triples with the given predicate. The
object of @rel (conversely the subject of @rev) is the "RDF identity" of
the current element...

- the RDF identity of an HTML element is, in order of precedence
 - @resource
 - @href
 - @src if it's an IMG
 - @about
 - a new bnode

meaning that we would get

<> a:b #b.

for the element above and if we added a @href then suddenly the @about
becomes the subject of a:b.

I am not saying this is not confusing, I am just saying that this is
_not_ related to the collection syntactic sugar. Ie, I still believe the
collection issue is, essentially, closed (I can use @resource in all my
examples, eg, as resource="_:", without problems).

I am not sure what to do with that stuff. I have to run now, maybe you
may want to look at the different cases with the @about removed from the
RDFI calculation.



[snip]
> 
> I have a visceral problem with about="_:", and that is that it makes
> bnodes explicit, which I really don't want to do to HTML authors. That's
> just too much RDF.
> 
> We may begin to hit diminishing ROI here, and I'm starting to lean
> towards supporting fewer of these constructs in order to not complicate
> the syntax. I can't see myself being convinced that about="_:" is going
> to help without hurting more....
> 

I do believe that (1) "_:" will be very rarely used and for RDF people
only but (2) in some edge cases it is difficult to avoid it. Any other
syntactic solution would lead to real complications I believe, and I
also believe that getting a sequence of anonymous and untyped bnodes is
sometimes necessary...

Ivan



> -Ben
> 

-- 

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