XHTML2 WG Weekly Teleconference -- 25 Apr 2007

-> http://www.w3.org/2007/04/18-xhtml-minutes Previous

<alessio> hi steven, shane, all...

Hi there Alessio

<yamx> I just joined in tel (Yam)

<alessio> idem, now

<alessio> i'm second

<scribe> Scribe: Steven

Announcements

Steven: There are currently: 11 participants from 8 organizations

2 Invited Experts

scribe: I know of 2 more organisations and one more invited expert in the pipeline

<Rich> what dates?

<yamx> OMA has a meeting in Bangkok during June 11-15. (Yam)

Steven: We have the option of meeting face-to-face 11-12 June
... at IBM in Hawthorne NY

<Rich> that does not interfere with my PF meetings in germany - a plus

<scribe> ACTION: Steven to ask WG by email if we should meet 11-12 June in NY [recorded in http://www.w3.org/2007/04/25-xhtml-minutes.html#action01]

<ShaneM> For the record, I can likely make that assuming nothing bad happens. Ill parents etc.

Steven: Otherwise we will have an editors meeting instead

Rich: Did the charter get updated?

Steven: As far as I know

<ShaneM> I nominate Steven as chair for this meeting.

Rich: Do we have a chair yet?

Steven: No, not yet

<yamx> I support the nomination (yam)

<ShaneM> is the current topic "charter"

Shane: For the record, the charter still talks about "The accesskey module"

class/rel/rev/role and Curies

http://lists.w3.org/Archives/Public/public-xhtml2/2007Apr/0023

Mark: I can see a big difference, and no one else can
... it's not so much role vs class, but should role be rdf:type
... I'm not sure if everyone has tried it against all use cases
... My next email at http://lists.w3.org/Archives/Public/public-xhtml2/2007Apr/0028.html
... discusses it more fully
... I know that Rich followed up saying that we want rdf:type
... but when Raman first suggested role
... his only example seemed to use role just as a hint
... so to repeat, I'm not sure that all examples have been tested against the idea that it represents rdf:type
... does @role give us a hook, where the rdfness is implied
... or does role give us a first-order rdf:type?

<ShaneM> is there another way to specify an rdf type in RDFa today?

Mark: We could further refine this
... just as we say this is what we mean by rel/rev etc
... we can say this with role
... at present we use class, but the suggestion is to use role instead
... and there is longhand <link rel="rdf:type"...
... the question is about the shorthand

Shane: I don't understand the subtleties
... but I am very worried about using @class because of the legacy
... on the other hand I understand the TF wants to take advantage of existing infrastructure

Mark: One discussion is whether we should use @class at all
... I believe we should, though I accept that not everyone agrees
... I expected a Dan Brickly to jump in and say "Ah! rdf:type this or that...", and that didn't happen
... the 2 use cases at present are 1) Rich's case, where rdf:type is handy
... 2) The mozilla/firefox one, where the element aquires the properties associated with that role
... these are the accessibility extensions

Rich: Role links to the rdf role taxonomy
... SAP, Google, IBM are using this stuff
... and even IE is going to implement role!
... (this is not official, but they really have to, to give access to the accessibility APIs)

<alessio_> yep :)

Mark: Will IE implement it in the same way?

Rich: It might

Mark: I'm trying to work out if it reflects on the role/rdf:type issue

Shane: I think it should be rdf:type

Rich: I agree

[Scribe got the last two wrong]

Shane: I think role = rdf:type is fine

<markbirbeck> Have we simply said that role == rdf:type? Have we just created a shorthand?

Mark: The problem is that if you make that decision once and for all, you can't undo it

Mark: better to define it per document

<markbirbeck> Something has a 'xh:role' of range...

Mark: and define a mapping if needed for a particular document

<Rich> http://www.w3.org/WAI/PF/Group/GUI/

Steven: What about <p role="rss:description".>
... what is the rdf equivalent?
... if we say it is per document, then I can't process that generically

Shane: That should be @property

Mark: No, because you need a predicate of type rdf:type
... because rss:description is not a predicate
... it is the object of a statement

<ShaneM> I thought this was more correct:

<ShaneM> <h2 property="rss:title">XFrames</h2>

<ShaneM> <p property="rss:description">An XML application for composing

<ShaneM> documents together</p>

Mark: What I am wary of
... if I have an item of type X
... will there be a situation where that item doesn't take on *all* the properties of an X
... in Raman's original example that was the case for instance
... so is rdf:type correct here?
... I'm scared we might have to invent a new attribute in a couple of years

Rich: Raman added role to a handler so he could tell what the handler was supposed to do

Mark: So you are saying this functionality can be got in a different way?

Shane: At the AOL FtF we talked about this
... and discussed the relationship between access/role and handlers
... for handling with navbars and so on
... Wasn't Raman there? Think so.
... Anyway, I think the concept of role has evolved since then

Mark: Sure,we said let's use QNames to extend it, and let you use it with rdf
... the problems are similar to those of @class

\... class is left very vague

scribe: should we leave role vague, or define it hard-and-fast

Shane: Surely the fact that is rdf:type is only relevant in the presence of other context?

Rich: role has multi-values
... doesn't that solve it?

<ShaneM> My brain hurts.

[Fast discussion about what rdf:type implies]

Mark: I mean that role and type are different, related but not equivalent
... I have a date of birth, doesn't mean I am of type date of birth

Steven: That's different
... if I have an element of role Mark Birbeck, then I would expect it to contain an element of role DoB

Mark: Don't confuse types with predicates

Steven: Top of hour; I think we have to just try to code up some examples and see what breaks

<Rich> <rdfs:comment>This is a draft resource for adding semantic

<Rich> annotations and information to Web resources. Knowing what the types of

<Rich> content are in a Web page allows for better customized renderings. Including

<Rich> want to allow information hiding, or rendering common types of content in a

<Rich> consistent way - it makes the Web easer to use and more accessibility. You

<Rich> can also use this for device independence so that content is better formatted

<Rich> to fit on the device. </rdfs:comment>

Steven: as a way of concluding something

<Rich> <rdfs:comment>You have to give each content type a title and

<Rich> description sorry folks (just be glad I did not put a min length on the

<Rich> description</rdfs:comment>

<Rich> http://www.w3.org/WAI/PF/Group/GUI/

<alessio_> I agree with Steven for concrete examples

ADJOURN

<scribe> ACTION: Steven to create some role use cases [recorded in http://www.w3.org/2007/04/25-xhtml-minutes.html#action02]

<markbirbeck> As I said, the problem is not in the mark-up, but in the triples generated, so it's not really a case of finding sample mark-up.

- DRAFT -

XHTML2 WG Weekly Teleconference

25 Apr 2007

Attendees

Contents

Announcements

class/rel/rev/role and Curies

Summary of Action Items

Scribe.perl diagnostic output