- From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
- Date: Sat, 14 Aug 2010 14:26:05 +0200
- To: Mark Birbeck <mark.birbeck@webbackplane.com>
- Cc: public-rdfa-wg@w3.org
Hi Mark, Mark Birbeck, Sat, 14 Aug 2010 07:33:42 +0100: > This is an interesting problem. > > I agree that if @longdesc has a 'meaning' in HTML 5 I don't believe it has "more" meaning in HTML5 than in HTML4 and XHTML. (At the moment, @longdesc is not permitted in HTML5 - but that is another, though not unrelated problem.) It is best, I think, to focus on HTML and XHTML in a "universal" way, I think. > then there should > be a way to indicate what that meaning is so that an RDFa parser can > pick it up. I don't think it would be good to put @longdesc into RDFa > Core though, because ultimately we want to provide a framework for > host language semantics, rather than providing those semantics > ourselves. > > So, what might work is to: > > 1. Add a new concept to RDFa Core that defines mappings from an > attribute name to a predicate (URI). These could be set by the host > language in just the same that terms are. This list of mappings is > then available to the parser and as traverses the tree it compares > each attribute against this list. I don't see any reason why this list > couldn't also be modified via profiles. Modification via profiles sounds meaningful. > (We'd also need to indicate in some way whether the attribute content > is a URI or a literal, but I'm sure we can work that out.) > > 2. Create some template of words that could be added to a host > language specification which makes it clear to implementers that some > particular attribute requires mapping during RDFa processing. > > This latter is going to be tricky given that RDFa is not actually > 'present' in HTML 5. I'm not certain what you mean by "present". Could it not be made clear to implementors via the HTML+RDFa specification? > But since other metadata solutions such as > Microdata would probably also want to map @longdesc, then the host > language merely needs to mention that @longdesc maps to a predicate > with a URI of <x>, and leave it at that. > > Do you think that this would capture what you want to see? Probably. :-) But do you have one or two small examples that show how it could be done? For example, does the solution(s) you suggest require that one also add @rel, @resource and @content to the <img> - as explained by Ivan? And are 1. (profile etc) and 2. (template etc) two different ways to achieve the goal, or are both 1. and 2. required to make it work? -- leif halvard silli
Received on Saturday, 14 August 2010 12:26:39 UTC