Re: [RDF/A] A non-XHTML Example

John,

Thanks for your detailed comments, again.

> Hmm, it sounds like you're not willing to go there yet -- I wish  
> you would be though! -- and that you're leaning towards removing  
> the statements altogether if a "generic XML application" presents  
> too many problems.

Well, we certainly believe that RDF/A in other XML dialects is a good  
idea, we just don't want to bite off more than we can chew (the task  
force is small). Our feeling (or maybe it's just *my* feeling) is  
that RDF/A should prove itself in XHTML before we claim that every  
other dialect should adopt it. We have done the best we can to ensure  
that this *can* happen at some point.

>  1. The primer/spec should clearly state its relationship with  
> Modular XHTML. I suspect the XHTML2 crew is creating an optional  
> module that contains the RDF/A attributes; therefore the RDF/A  
> attributes can be used in XHTML Family document types other than  
> just XHTML2. If not, then I'd suggest you contact them to ensure it  
> does perhaps requesting that it be named "RDFA" or something. It's  
> fortunate for RDF/A that the id and href attributes are required in  
> all Modular XHTML doctypes.

Yes, the XHTML2 WG is definitely creating this as a module, and I  
suspect they will explain this clearly in their specification. The  
Primer should probably make it clear that while I'm on the SWBPD WG,  
Mark Birbeck (co-author) is on the XHTML2 WG, so this is definitely  
joint work.

>  2. It's not clear that RDF/A attributes are used on XHTML2  
> namespace elements, regardless of the type of XML entity in which  
> they are located. In other words, the XHTML2 elements are not  
> necessarily located in XHTML2 entities; they can be located in any  
> XML entity. I suggest an example of this technique would be useful.
>
> <foo:Document>
>       <html:section about='url'>
>             <span property='dc:title foo:FullName'>Jon Doe</span>
>       </html:section>
> </foo:Document>
>
> Please note that I still assume that the 'property' attribute is  
> defined as QNAMES, not just QNAME, in the module. If that is not to  
> be, please tell me why not. If it is, let me know so that I can  
> stop asking about it.

This is an interesting point we have not considered, whether property  
can include multiple possibilities. We'll take it up in the TF.

> 3. An XHTML2-oriented guideline may have more gravitas if it were  
> co-authored with the XHTML2 WG. Within the text, I suggest readers  
> be directed to relevant information in the XHTML2 specification  
> and, for the RDF/A spec, you likely need to indicate which spec has  
> precedence -- yours or the XHTML2 spec -- and in what matters.

So, again, just to be clear, all of this work is co-authored with the  
XHTML2 WG. Mark Birbeck and Steven Pemberton are on the TF. We should  
make that much clearer, as we certainly could not have done this work  
without their help.

-Ben