W3C home > Mailing lists > Public > public-webid@w3.org > October 2012

Re: Comments on current editor's draft (Web 1.0, 3rd October 2012)

From: Antoine Zimmermann <antoine.zimmermann@emse.fr>
Date: Wed, 24 Oct 2012 09:05:03 +0200
Message-ID: <5087931F.7010602@emse.fr>
To: Henry Story <henry.story@bblfish.net>
CC: "public-webid@w3.org" <public-webid@w3.org>
Le 24/10/2012 06:11, Henry Story a écrit :
>> 4.  I don't see where the use of Bob and Alice actually helps. A
>> generic "user" or "user agent" would be fine, as far as I can see,
>> since we don't rely on the notion so heavily;
> I'd have to look at it, but I thought that having names made the text
> lighter. That is it avoid having to use the word Agent all the time,
> which becomes grating. But perhaps you have an example where we
> overdid it?

On the contrary, you don't do it as promised. You say "we will say Bob 
wherever we talk about the subject" but most of the time, you say "the 
subject". The use of "subject" or "agent" is not shocking and is 
natural, so in the end, I don't see Bob and Alice as being very useful. 
To make things clearer, I think it is better to use examples, where you 
can have Bob and Alice, rather than explaining the concepts themselves 
with Bob and Alice.

>> Sec. "the query engine MUST support the D-entailment regime
>> fpr xsd:hexBinary" -> this implies that the query engine MUST
>> support RDFS entailment, since D-entailment subsumes RDFS
>> entailement. This is unlikely to be the case.
> yes, which is why I added the algorithm in the next section. I don't
> think we can really add that it is unlikely to be the case in a
> spec...

In the RDF 1.1 WG, Richard Cyganiak emmitted the idea of a light-weight 
inference regime where only datatype inferences hold.
Take a look at this thread: 

However, this has not been discussed in meetings and telecons, so don't 
rely on it for the moment.

What I'm wondering is why you impose such a strong constraint on the 
query engine while, IMO, simply normalising xsd:hexBinary would be 
sufficient as a MUST, and RDFS- or D-entailment would be a SHOULD.

In any case, nowhere in the examples or definitions appears a need for 
inferences, so this conformance constraint comes as a surprise and seems 

>> C. References: [RDF-SPARQL-QUERY] -> consider reference to SPARQL
>> 1.1 (not yet standardised but quite stable already) Why is there a
>> referencec to RDFa 1.0 and to RDFa 1.1, both for the formal
>> syntaxes and the primers? [TURTLE-TR] -> should use RDF 1.1 Turtle
> that just came out :-)
> should we also update to latest RDFa?

Latest RDFa, yes because it is already a recommendation. For Turtle, you 
can prefer the Last Call version, certainly in good way to become a Rec 
soon, because the older Turtle was not a recommendation anyway.
For SPARQL, the situation is different, SPARQL 1.0 being a Rec as 
opposed to SPARQL 1.1 where there are still some issues to be solved. 
But you can probably cite 1.0 as normative and inform the reader that 
there is 1.1, with a lot of supporting implementations.

Antoine Zimmermann
ISCOD / LSTI - Institut Henri Fayol
École Nationale Supérieure des Mines de Saint-Étienne
158 cours Fauriel
42023 Saint-Étienne Cedex 2
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
Received on Wednesday, 24 October 2012 07:06:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:54:37 UTC