See also: IRC log
ACTION: Ben to reconstruct action on role attribute with help of Danbri [recorded in http://www.w3.org/2005/09/27-swbp-irc#T14-28-08] [DONE]
-> The ROLE attribute [Ben 2005-10-02]
Ben: Lisa Seeman has asked to participate in a
discussion of Role next week
... please look at Lisa's
document for next week
-> Re: The ROLE attribute [Lisa 2005-10-03]
Ben: main question is whether role is syntax for rdf:type or we need a separate xhtml2:role that is rdfs:subProperty of rdf:type
Steven: some examples in the wiki lead me to suggest that we could use the role attribute as a shorthand for rdf:type
Mark: we should consider whether role has any
relation to rdf:type
... one of the examples had lots of rdf:type properties so Steven suggested
this shorthand might be nice
... but I am having second thoughts
... something that plays the role of something else is not necessarily of
that type
... e.g. a toolbar can play the role of a footer but not _be_ a footer
... saying a toolbar is a footer might attach all sorts of other properties
by inference that it might not legitimately have
Ralph: yes, the client application might want to be able to distinguish between rdf:type footer and things of other types that simply have that role
-> discussion of role in meeting record of 2005-08-02
-> discussion of role in meeting record of 2005-07-26
<Steven> Quote: Steven: the danger of the
id solution is that id plays so many roles that you could accidentally insert
an id (to be a link target) and suddenly change your RDF
... in the rdf:type example, I would have written <section
role="foaf:Person">
... I see Mark doesn't use role much, whereas I use it a lot
... why this difference in approach?
... maybe role is being though of differently by each of us and we should
write down what role means
[Steven is quoting from 26-swbp-minutes]
Ralph: can we describe what 'role' does mean?
Mark: maybe 'role' is the wrong word
... in a discussion of several years ago, Raman said that 'class' would be a
better word but that word was too confusing w.r.t. CSS
... I explain 'role' to people more in terms of 'intent' or 'purpose'
... i.e. "the purpose of this section ..."
... Raman's examples were to explain why a <script> element was
included
... e.g. the javascript in a <script> plays the role of a hint
... Raman's use cases imagined a server that adds "role='hint'" to legacy
scripts that can no longer be interpreted, then a client can substitute some
other hint
... i.e. the intention of this block is X and I know how to do X some other
way so I can substitute
Steven: I'm not sure these definitions clash
... seems ok to say that something that has a role of a navigation element
also has that type
... [even though] a foaf:Person does have more semantics than a navigation
element
Mark: make sure we're not closing off some
avenues, particularly with the work that Lisa is doing
... make sure we're not going to cause something to acquire lots of
inappropriate properties
... perhaps we need another attribute
Ralph: I'm reluctant to start down the path of additional syntactic sugar for common properties; where will that end?
Mark: would like more feedback from RDF experts on whether the semantics of role is consistent with Class
Ralph: I need help from HTML experts to understand what role is intended to mean
Mark: in the early days we considered role to be like a hint; related to appearance
Steven: yes, but then we expanded its use to
provide a facility to avoid having to add lots of new elements
... e.g. <span role="person"> adds semantics to the span
Mark: should elements that do add semantics
also generate an rdf:type triple?
... e.g. is <div role="address"> the same as an <address>
element?
Steven: address is one of those few elements
that do carry an aura of semantics
... title is a better example; we have said that <title> is the same as
<meta property='title'>
Ben: do we all agree that there is some
semantics here that could be as light as xhtml2:role or as strong as
rdf:type?
... we're trying to figure out how strong a statement we're making with
role
Steven: yes; it's clear that role does specify semantic information. We're trying to clarify its role w.r.t. RDF
ACTION: Ben summarize the question of semantics of the HTML role attribute and solicit feedback from RDF experts [recorded in http://www.w3.org/2005/10/04-swbp-minutes.html#action01]
Ben: my concern is not href with CURI -- people who use that construct need to know that old browsers won't recognize CURIs; the bigger concern is other attributes like rel that use qnames
Mark: I think it's OK; if CURI is a superset of
QName then if QName works in rel a CURI will also work
... e.g. in XHTML2 we're specifying that role='next' is shorthand for
'role=xhtml:next'
Ben: aren't we requiring additional syntax for CURIs?
Mark: only to resolve the ambiguity in the
datatype that accepts both URIs and CURIs
... the tricky bit is that we have one syntax for rel (only CURI) and another
syntax for href (either URI or CURI)
Ben: so rel='cc:license' would be ok but rel='http://creativecommons.org/license' is not ok?
Mark: right
... two different syntax for rel and href -- it would be nice if there were
only one syntax but that creates a backwards compatibilty issue
Ben: if there were some magic that let old browsers understand CURIs in href, we'd want to use that magic
Ralph: two syntaxes is more cognitive load for users but compatibility gets higher priority for me
Mark: with what are we trying to maintain
compatibility?
... what exactly would be the issue with "rel='[next]'"
... we've already said that existing XHTML1 documents might not be XHTML2
documents
Ralph: I think that's going to be a big issue (XHTML1 doc not being a valid XHTML2 doc)
Ben: summarizing current proposal; rel
is CURI only and about, href are either URI or CURI
... so change from earlier discussions is that rel, rev,
property, role are upgraded to CURI
Ralph: I suggest polling the XHTML WG and SWBPD
WG more formally
... I have just polled the W3C Staff about this CURI idea
Steven: HTML WG has already discussed; Discussion of Qnames and URIs [Member only]
Mark: I am writing a document summarizing
CURIs
... IPTC has already adopted this as a solution
... they had initially wanted W3C to change the definition of QName
ACTION: Mark write CURI specification by 10 Oct [recorded in http://www.w3.org/2005/10/04-swbp-minutes.html#action02]
Ben: see my notes in Mark, Ralph, and Ben meet up [Ben 2005-09-29]
Ben: let's table this until next week when I
hope to have more examples to flesh out these inheritance ideas
... let's also consider Jeremy's
concerns about reification
-> CC use case [Jeremy 2005-10-04]
Ralph: I think reification plays an important role in helping people use RDF but it certainly has been confusing to users and there is considerable opinion that it should be dropped from RDF
Mark: Reuters feels that reification is very
important
... provenance; who has made a statement, when they said
it, etc.
Ralph: one of the original use cases for
reification was to support PICS; see PICS Rating Vocabularies in
XML/RDF
... I'd like to see other big use cases
ACTION: Jeremy consider Creative Commons 'who made this license' use case w.r.t. reification [DONE] [recorded in http://www.w3.org/2005/10/04-swbp-minutes.html#action04]
Ben: Jeremy's example is not something that Creative Commons would adopt in the way he proposes as it changes too much
ACTION: Ralph and Ben to augment the issues list [recorded in http://www.w3.org/2005/09/27-swbp-irc#T14-30-04] [CONTINUES]
<Ralph> I claim some progress
ACTION: Ben to put together the "ACID" test for XHTML2 RDF/A [recorded in http://www.w3.org/2005/07/26-swbp-minutes.html#action02] [CONTINUES]
ACTION: Mark to check edge cases of inheritance in RDF/A [recorded in http://www.w3.org/2005/07/26-swbp-minutes.html#action06] [CONTINUES]
[adjourned]
Change History:
$Log: 04-swbp-minutes.html,v $ Revision 1.2 2005/10/04 15:51:49 swick Cleanup for publication