Re: ISSUE-35 RDFinXHTML-35 (also ACTION-240)

On Apr 21, 2009, at 12:17, Harry Halpin wrote:
> So, instead of saying "Let's reject CURIEs", what can constructively
> can be done in the opinion of people from HTML5/WHATWG to fix CURIEs
> so RDFa plays well with HTML5?

The simplest fix is to use absolute IRIs instead of CURIEs.

> Is there anything outside of CURIEs that is found objectionable in  
> RDFa?


Many times RDFa seems to be a solution in search of a problem. That  
is, it starts with the assumption that it should be able to express  
arbitrary RDF graphs instead of starting with use cases. I think the  
result is that RDFa is not a good syntax for any particular use case.

Using URIs as identifiers is a problem--not a solution. Namespaces in  
XML, CURIEs and the pattern that the RDF community abandons one syntax  
after another to try out a new serialization all flow out of trying to  
hide the unwieldy length of URIs by syntactic tricks.

(Then there are the non-technical objections to the flagship use case  
of RDFa--the CC+ use case--based on Free Software / Open Source  

> If this discussion has
> already happened elsewhere (I imagine it has), and there has been some
> consensus from non-RDFa techies about how to fix bits of RDFa to make
> it better, I think this should be taken very seriously.

Hixie is currently developing a proposal that approaches the stated  
use cases of embedded microdata from the point of view of use cases as  
opposed to approaching it from overlaying the RDF data model onto HTML.

In the interest of time, I'm going to refrain from putting forward  
alternative proposals until Hixie's proposal is available for review.  
(Scheduled to happen later this month.)

> I just spent the last few days at a conference on information
> retrieval with researchers from Yahoo!, and Yahoo! is producing
> applications that consume RDFa, and now I'm seeing lots more sites
> produce it, although not in XHTML, but in something resembling HTML5.
> I think Yahoo! has stats on how users actually deploy it, and it would
> be great if this could inform the discussion.

My take on reports of Yahoo!'s implementation is that RDFa is too hard  
for implementors, authors and documentation writers.

Henri Sivonen

Received on Tuesday, 21 April 2009 12:12:10 UTC