RDF in XHTML TF

CURIE and IPTC

<benadida> http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2005Oct/0092.html

Mark: I think the IPTC metadata will mostly be transported in newsml "packets"
... in an (XML) envelope
... not in the HTML payload

Ben: CURIEs did two things for us: (1) they allowed abbreviation of URIs like QNames but without restriction on the 'name' part and (2) they were a convenient solution to the bnode reference issue

Mark: note that Misha said that IPTC intends to use CURIE regardless of what XHTML2 does

Ben: CURIE with []?

Steven: IPTC is defining its own attributes, so they don't need the '[]' approach

Ralph: the team hesitates to have another mechanism for referencing
... why not use entities if abbreviation is a big issue
... the bnode question hasn't bubbled up yet
... general unease about new syntax (square brackets in particular)
... any new syntax would raise concerns; it's not specifically about our choice of square brackets

Jeremy: several concerns; partly about the length of identifiers, partly about bnode references
... the CURIE solution provides a way to not have to put all the complexity up front

Ralph: perhaps if we had presented the requirements and the pros and cons of each of the options then the rationale for the CURIE solution would be better understood

Ben: several of Mark's mails have the pros and cons of options we considered
... I'd like to start preparing a [rational] document next week

Ralph: is the issue that xmlns doesn't work the way IPTC wants a real issue for us?

Jeremy: yes, xmlns doesn't work the way anyone wants
... I've found several cases that will cause problems; e.g. embedding XHTML2 inside an [RDF/A] XHTML2 chunk

Mark: IPTC is talking about using some other attribute than xmlns
... to declare the substitution rules
... might even permit the declarations to be contained in a separate document

Jeremy: we've gone to CURIE to abandon syntactic restrictions on the right-hand side of QNames
... perhaps we can abandon the left-hand side of QName as well

Mark: CURIE could define substitution mechanism without specifying the source of those substitutions
... even XPath is using namespace prefixes; no one has proposed entities
... I've blogged about the CURIE solution last week and folks are commenting positively

Steven: Mark's blog entry was picked up in the O'Reilly Developer Weblogs

issue 7: syntactic sugar for class attribute

<benadida> syntactic sugar for class attribute

Ralph: concerns about overloading of the class attribute
... the real use of class is to make CSS work, right?

Steven: why do we need to do this if we use role? role and class are very similar but role has more clear semantics
... role is meant to do the right thing

Steven: class is now also used for microformats

Mark: when role was first proposed it was because class would be overloaded
... but the reason class would be overloaded was that it did not use QNames
... now class has been extended to define some semantic meaning but people feel class and role have different semantics
... it would be a mistake to not define a meaning for class

Steven: this would be a big change for the entire Web community due to namespace defaulting
... people would not be able to use unqualified names in the way the used to use them

Mark: if I want a foaf:address and also to style these in green I shouldn't have to duplicate more markup

Steven: an unqualified name is supposed to be interpreted as in the local namespace

Ben: we could define class to be a CURIE

Mark: we could also define a different namespace defaulting rule

Ralph: it would be too confusing to have different namespace defaulting rules

<benadida> class="foo"

Mark: we've defined the namespace rule to make rel='next' work the way we want, with next in a particular namespace

Ben: if we want class="foo" to mean foo in a local namespace then we either have a 3rd namespace defaulting rule or class is a URI only, not a CURIE

Mark: alternatively we can say that unqualfied value does not have a defined namespace
... e.g. the union attribute [in XML Schema] is defined to be based on the target namespace

Ben: next meeting 8 Nov?

Steven, Jeremy: regrets

Ben: please use mailing list for additional comments

$Log: 01-swbp-minutes.html,v $
Revision 1.2  2005/11/01 21:19:23  swick
Cleanup for first publication