RDFa in XHTML Task Force

Action Items

<manu> http://www.w3.org/2010/01/21-rdfa-minutes.html

<manu> ACTION: Manu to create basic RDFa API document. [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action01]

<manu> ACTION: Manu to test CORS support for Pre-loading prefixes in latest browsers. [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action02]

<scribe> ACTION: Manu to contact Dan Brickley to discuss Microdata to RDFa mapping. [DONE] [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action03]

Manu: Dan Brickley has been talking to Hixie on WHATWG mailing-list to see what the overlap between Microdata and RDFa is.

There seems to be a feeling that Microdata is at the very least a subset of RDFa.

It may be possible to create a profile to use XSLT to map from one to the other.

<scribe> ACTION: Manu to get in touch with LibXML developers about TC 142. [CONTINUES] [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action04]

<scribe> ACTION: Mark to generate spec text for @token and @prefix [CONTINUES] [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action06]

<scribe> ACTION: Mark to generate spec text for pulling in external vocabulary documents [CONTINUES] [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action07]

<scribe> ACTION: ShaneM to identify the requirements for html2ps and see about getting reSpec to support them. [CONTINUES] [recorded in http://www.w3.org/2010/01/28-rdfa-minutes.html#action08]

<ivan> http://www.w3.org/2010/02/rdfa/

RDFa Community Updates

Manu: Had a chance to read Benji's comments?
... Benji wrote blog-post that was quite critical, some of it unfairly, of RDFa community.
... He feels that pushing Microdata is a better option for adoption.
... His past posts of the years are along much the same lines, that the solution should be a collection of tokens, a la Microformats.
... We should take his comments seriously, since he represents an identifiable set of concerns.

Ivan: We did discuss the possibility of defining Microdata as a profile of RDFa, and I would be in favour of that.
... But this post talks of removing things that it doesn't seem possible to remove, such as @resource.
... He doesn't really give an argument for removal.
... Same goes for ranting against CURIEs -- we're working on that and there's not much more we can say.
... The idea that one set of attribute names is better than another is really a matter of faith.
... Whilst we don't care that much about the attribute names, a big problem is existing deployments.

Steven: I've been working on RDFa for at least 7 years, and it's not as if it suddenly popped out of nowhere.
... It had very thorough discussion, use-cases, etc.
... Microdata comes out of nowhere, but somehow it's presented that RDFa-proponents are in the wrong.
... Appreciate the work of bridge-building, but we have to remember that no-one was asking for Microdata.

Mark: I missed the call on Microdata profiles - I think I agree with Steven, I'm not sure I understand what the benefit is.

Mark: We've been working on this for years and Microdata is in flux - and is progressively getting closer and closer to RDFa.

Mark: I don't mind the dialog, but we risk damaging deployment if we change all the attribute names.

Mark: It's sort of religious - one attribute is better than another.

Mark: but there is another way to view this.

Mark: Changing rel to itemprop may be a bad change - @rel and @rev already has a place in HTML.

Mark: We've been very careful to re-use pre-existing semantic elements in HTML.

Mark: Microdata duplicates @rel and @rev - so there is a problem here - that new attributes are being defined for old concepts.

Ivan: Changing attribute names is out of the question.
... On the other hand I don't mind about building bridges.

<Zakim> Manu, you wanted to discuss harm that continued bike-shedding may cause.

Manu: I'm increasingly concerned about discussions with people like Wikipedia.
... The problem was that although we know the background to Microdata popping up from nowhere, the outside world doesn't.
... They just see a 'fight' going on.
... I think it's harming adoption.
... If others implement a Microdata API in the browser, then that could affect adoption.

Mark: I don't know if a GRDDL-approach, mapping Microdata to RDFa would work... we've had that discussion before.

Mark: I think we should focus on RDFa API and getting RDFa out there more.

<ShaneM> I don't think the transform can be isomorphic.

Mark: RDFa and Microdata are fundamentally not equal.

Mark: This is the same thing that happened with Microformats - RDFa kept going and I think Microformats are coming to an end soon.

Ivan: Wiki would be a good place to start on some of these ideas.
... (That was response to Mark saying that we should get on and document some of our ideas on RDFa API.)

<manu> +1

Ivan: If as Mark says, that RDFa is superior to Microdata, then we should be explaining this.

<ShaneM> pre-defined prefixes simplifies immediately

Ivan: If someone doesn't want the full functionality of RDFa it should be clear how to use it.

Shane would like it minuted that he is paying attention.

<manu> http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2010Jan/0059.html

RDFa prefix pre-loading

Ivan: At a theoretical and technical level, these things are not totally unrelated.

<ShaneM> +1

Ivan: They shouldn't need to be presented as being distinct.

Mark: One of the things on the vocab front is this gradual rejection that follow-your-nose is everything.

Mark: The idea that everything must be able to be worked out on the fly by default - you need to support follow-your-nose

Mark: But saying that stuff is predefined isn't necessarily a bad thing - we shouldn't be scared of doing that.

Mark: This is where Microdata and Microformats falls down - they don't have the follow-your-nose mechanism.

Mark: RDFa does.

Mark: as long as you can ensure that the two co-exist, then why not pre-load prefixes?

<manu> Ivan: Then we largely agree, right?

Mark: The downside to this follow-your-nose is that it is used as an argument against pre-defined vocabs.

<manu> Ivan: Google uses the same mechanism... but it has been said all along that a processor would cache a bunch of these namespaces.

<manu> Ivan: caching is a fundamental part of the web.

Ivan: Not disimilar to an editor caching DTDs.

<ShaneM> I want to do this RIGHT NOW

Manu: But which vocabularies would we put into the spec? Which ones have longevity?
... Top 20 from prefixcc?

Mark: I think we shouldn't be /scared of doing this/ - not /do this now/.

Mark: We should certainly think all of this stuff through... there isn't a principle that we shouldn't have to pre-define things.

<html>

<head profile="http://rdf.data-vocabulary.org/profile/2009">

scribe:

</head>

<body>

<div typeof="Person">

<span property="name">John Smith</span>

<span property="nickname">Smithy</span>

<span property="url">http://www.example.com</span>

<span property="affiliation">ACME</span>

<span rel="address">

<span property="locality">Albuquerque</span>

</span>

<span property="title">Engineer</span>

<a href="http://darryl-blog.example.com/" rel="friend">Darryl</a>

</div>

</body>

</html>

Mark: This approach might be better - since it's encapsulated in a @profile attribute.

Mark: That is better than Microformats and it's scalable.

Mark: This would allow RDFa processors to understand the URL to mean something special.

Mark: It's a key to tell the authors that their data would be processed in a particular way.

Mark: Maybe we should defer the pre-loaded prefixes until we can have this discussion

Mark: Not intending to take the discussion in other direction, just trying to point out that there may be something even better than fixing prefix values.

<Zakim> ShaneM, you wanted to talk about the process for picking vocabs

Shane: But people forget to use @profile, or they don't have access to <head>, etc.

<manu> +1 to Shane's comment

Shane: So having the ability to use "dc" or "foaf" without having to define it, would be very useful.
... Let's just do this now.

<ShaneM> rule one for picking prefixes. vocabs defined by the W3C are predefined.

Mark: Note that omitting @profile is a red herring.

<manu> how so?

Mark: People raise it, and say that no-one ever used @profile in Microformats.

<manu> ahh, yes

<manu> I see your point

Mark: But in RDFa if you miss out all the namespaces you're scuppered anyway.

<manu> The reason they didn't use it was because it had no effect on processing.

Mark: So there's already a basic requirement on authors to add something.

Mark: Yes, exactly.

Mark: But if you miss out the Google namespace in your RDFa, you won't get indexed properly.

Mark: Same in Yahoo!.

Mark: So all we're saying is instead of adding 10 URIs, just add one.

<manu> The #seo people will definitely make a point of it

Mark: A point of getting it right?

<manu> yes

Mark: Exactly.

Mark: Anyone interested in providing information for some consumer will want to get it right.

<ShaneM> but....

Mark: Control over <head> is a fair point though.

Mark: Shane is right there.

<ShaneM> profile goes on the html element. or the head. I can never remember. xmlns can go anywhere. and we had a requirement that you need to be able to do this stuff anywhere.

<manu> true

<ShaneM> thanks

Mark: Yes, I agree with you there.

Mark: But we also discussed having <div profile="...">

Mark: That would be a nice feature anyway.