Re: ISSUE-15 PROPOSAL: @version attribute in HTML5

On Jul 8, 2010, at 4:02 AM, Ivan Herman wrote:


On Jul 8, 2010, at 06:35 , Manu Sporny wrote:
[snip]

We have a couple of options in front of us, that could be combined:

1. Keep everything as it is right now, ignoring the polyglot
 ramifications of the decision.
2. Harmonize the HTML5+RDFa and XHTML5+RDFa @version strings.
3. Specify that @version SHOULD NOT be used unless an author
 would like to force the RDFa Processor into a particular mode.

I think that we should do #2 and #3:

+1


We should harmonize HTML5+RDFa and XHTML5+RDFa @version strings to:
 version="HTML+RDFa 1.1"

+1


We should specify that @version SHOULD NOT be used unless an author
would like to force the RDFa Processor into a particular mode. If
@version is not specified, the RDFa Processor should assume the latest
REC version (which will be RDFa 1.1 for the foreseeable future).

+1

What worries me in this formulation is that this sounds like RDFa processor are supposed to maintain an 1.0 as well as an 1.1 version. Do we really require that? At the moment, my experimental distiller is certainly RDFa 1.1, and I do not have any internal branch that would disallow @profile and such...

I recently updated my parser to be both RDFa 1.0 and RDFa 1.1 conformant. It did not prove to be that difficult:

For RDFa 1.0:


 *   Ignore @vocab and @profile
 *   Don't force xmlns: prefix to lower case
 *   Don't force CURIE prefix to lower case
 *   When extracting a URL, pass options on acceptable forms of URL (e.g.: :safe_curie, :uri, :term, etc). In 1.0, this is different depending on the context to only consider transformations consistent with 1.0 rules, basically a set of restrictions that are considered for different steps of the transformation, with  no restrictions imposed for 1.1

I haven't yet pushed a new version of the parser that does this yet.

Also, in my fork of rdfa-test-suite, I created an xhtml11-manifest, which changes some no-longer relevant tests to "onhold" status, and had previously added some based upon my on and Ivan's work. I'll probably add some more as I update to the latest draft, all "unreviewed", of course.

Ivan

Gregg