Re: How are default profiles normatively established?

On 24 Aug 2010, at 08:45, Ivan Herman wrote:
>> I can't quite tell what the model for default profiles is.
>>
>> Is it essentially a normative external profile document that is in  
>> effect for the entire document tree?
>
> That is my understanding...
>
>> Or is it essentially a set of mappings that are normatively defined  
>> in the host language's specification document, and SHOULD also be  
>> available in an informative profile document, purely for consistency?
>
> *Conceptually* the answer is (in my view) the previous, ie, default  
> term settings for a host language does not require any new/different  
> mechanism from what we already have. In practice, of course,  
> implementations may hardwire the default profile content in their  
> parsing because there is also a note somewhere in the text that  
> default profile specifications are not supposed to change, ie,  
> processors are advised to cache them somehow. (Sorry not to refer to  
> a particular place, I have a very bad connection today due to some  
> networking change at my home.)
>
> Whether a host language specification documents the content of the  
> default profile becomes an editorial issue.

What you are saying above makes sense to me, and as you say it is  
supported by some language in the spec. But there is other text that  
contradicts you. Section 7.2 states, when discussing term mappings and  
the default vocabulary:

“If a Host Language provides an initial list [of term mappings], it  
SHOULD do so via an RDFa Profile document.”
“If a Host Language defines an initial setting [for the default  
vocabulary], it SHOULD do so via an RDFa Profile document.”

This suggests that there is another normative mechanism, besides  
profile documents, for establishing term mappings and default  
vocabulary, and that profile documents are optional even if defaults  
are provided.

>> - What happens if the default profile document is unavailable due  
>> to server outage?
>
> If the caching mechanism works the processor should be able to go on  
> happily...

Ok.

>> - Is the default profile in effect while processing *another  
>> profile document*?
>
> I think the answer is a clear no. Any effect of profiling is local  
> to the document that refers to it; any profile document is an RDF(a)  
> document on its own right, completely independent of the caller. Ie,  
> I may process an SVG document with RDFa, using a default SVG profile  
> for the SVG part, but referring to an external profile document in  
> XHTML+RDFa. Clearly, the SVG default profile should have no effect  
> on the XHTML+RDFa one.

I apologize, I expressed myself poorly. I agree with the above. But my  
question was this. Let's assume, for the sake of argument, that  
HTML5+RDFa had a default term mapping that maps, say, “term”, “prefix”  
and “vocab” to their respective URIs in the RDFa vocabulary (rdfa:term  
etc).

Now let's say we process an RDFa document (whatever the host language)  
that contains a @profile reference. The profile document is an  
HTML5+RDFa document. When processing that HTML5+RDFa document, would  
the term mappings for “term”, “prefix” and “vocab” be in effect?

The exception at the end of the RDFa Profiles section made this a bit  
ambiguous, because it stated that only one “level” of profiles are  
possible. But as per your other email, you are considering to drop  
that exception, and that would my question moot.

Best,
Richard



> I would think that is already clear from the document *if* the  
> default profile is just a profile, ie, does not rely on any extra  
> mechanism.
>
>> I submit that the confusion has two sources:
>>
>> 1. The terms “RDFa Profile” and “RDFa Profile Document” are used  
>> interchangeably, making it unclear what a “Default RDFa Profile”  
>> is. (My intuition is that a profile is an abstract data structure,  
>> while a profile document is a way of establishing a profile.)
>>
>> 2. The section that explains RDFa Profile Documents does not  
>> explain how these documents relate to default profiles. (I think it  
>> should state that host languages SHOULD provide a profile document  
>> for their default profile, but that the profile document is never  
>> used in parsing.)
>>
>
> Something like that is probably necessary, you're right.
>
> Ivan
>
>
>> Best,
>> Richard
>
>
> ----
> 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
>
>
>
>
>

Received on Tuesday, 24 August 2010 12:36:03 UTC