Re: Request to publish HTML+RDFa (draft 3) as FPWD

Manu Sporny wrote:
> The 3rd draft of the HTML+RDFa specification has been released and is
> available here:

An additional comment:

Section 5.1 says:

   "All RDFa attributes and valid values (including CURIEs), as listed 
in Section 2.1: The RDFa Attributes, are conforming when used in an 
HTML5 or XHTML5 document." lists:

   rel, rev, content, href, src,
   about, property, resource, datatype, typeof

I don't see any restriction on what elements these attributes may occur 
on, so it sounds like they're being allowed everywhere.

That seems very bad, particularly for the attributes that are already in 
HTML and have some associated behaviour and DOM bindings etc. E.g. if 
href is allowed everywhere, someone will write:

   <ul><li href="about.html">Click me!</li></ul>

and the HTML5(+RDFa) validator will say it's perfectly valid, but it 
won't work like the author expects, so they will get confused. 
Similarly, elements that support the rel attribute in HTML5 have .rel 
and .relList properties in the DOM; if rel was allowed on every element, 
then only a random subset of elements that allow rel will have the 
associated DOM bindings.

So I believe these attributes (rel, rev, content, href, src) should only 
be permitted on the elements that HTML5 currently permits them on.

For the new attributes, it seems sensible to allow them on all HTML 
elements (i.e. elements in the HTML namespace), but I don't see any 
restriction like that, so it sounds like they're being allowed on any 
element at all. E.g. if I have inline SVG in an HTML5/XHTML5 document, 
it sounds like the attributes are being permitted on the SVG elements too.

That seems to be overstepping the jurisdiction of HTML5, in a way that 
HTML5 currently avoids doing - the conformance of SVG subtrees should be 
deferred to the SVG specs. So HTML+RDFa should only permit RDFa 
attributes on HTML elements, and not claim anything about non-HTML elements.

(This is all independent of the question of whether RDFa processors 
should process these attributes regardless of what element they're on 
(I'm not sure I want to ask that question yet) - it's simply about what 
is a conformance error for authors.)

Philip Taylor

Received on Wednesday, 23 September 2009 08:53:24 UTC