See also: IRC log
[Ben and Ralph talk about microformats while waiting for others]
Mark: I met again today with
IPTC, followup from Friday meeting; they're keen to progress
further
...
Friday meeting was very positive
... Misha focussed the discussion on his document first, then
we dived into details
... IPTC wants a small set of relationships between data
... they want to make statements about the relationships
between other tags
... reification is a big issue for them; they want to give
[provenance] -- who assigned a 'tag', with what confidence,
etc.
... they're very open to RDF/A, as long as they get a compact
syntax
... in their proposed syntax there is a lot of
level-mixing
... e.g. attributes whose subject is a statement and other
attributes whose subject is something else
... they admitted this could get confusing to people
... this can help us, as it's a real meaty application; I felt
very positive about this
... they have a tight timescale; their next meeting is in
October and they need to publish documents beforehand, thus 3
Oct is their deadline
Ben: how will they do RDF/A without XHTML2?
Mark: they have several
languages;
... one is comparable to XHTML2; it marks up a news story
... they're not completely convinced that XHTML2 gives them
everything they need
... they have a suite of other languages; e.g. SportsML that
sit on their base language
... then they have another layer that provides alternative
formats for a document
... a wrapper describes the 'package' of formats
available
... document format to wrap other documents
... another format: many documents that make up a news
story
... 3 document formats. XHTML2 might replace the bottom one,
but not others.
... RDF/A was designed as general purpose attribute syntax
Ben: can RDF/A be added to XHTML1
Mark: there's been discussion
about this.
... there are ways to mark up DanBri's examples using
XHTML1
Ben: very interesting to think about this issue. Could it be accomplished with XHTML1 that renders correctly in the browser?
Mark: property is not an XHTML1 attribute, but it could be added as a module.
Ben: how is XHTML1 modularization going to interact with validators?
Mark: there are probably no XHTML
1.1 validators, because there is no schema yet
... XHTML 1.1 will have modularization, in order to replace
XHTML 1.0.
... DTDs for XHTML 1.0 will be replaced by XHTML 1.1
schemas
... tried using XHTML 1.1 architecture to combine multiple
modules and add XForms. It didn't work.
... Xforms is a bit more complicated of a use case.
... currently tidying up XHTML 1.1 modularization. Later XHTML2
using same techniques.
Ben: keep us posted on this; it might help for adoption
Mark: tricky thing is QNAME issue
for predicates.
... could argue that there's nothing to stop you using Qnames
in REL.
... same as "DC.creator"
Ben: what's the current direction on qnames everywhere?
Mark: would be great to have
qnames and URIs interchangeable. square brackets won't fly
either.
... need to add qabout and qhref
... or qcontent
Ben: in terms of consistency, if we add qabout and qhref do we also need qrel?
Mark: yes
... [even though] it would be painful for every attribute to
have a q-version
Ben: if we do settle on the qabout, qhref direction then adding qrel would settle the problem
Mark: and existing rel would be a
kind of local identifier
... this could provide a neat answer to the backwards
compatibility issue
... might also be solvable by putting constraints on the
namespace prefixes; e.g. insist that the namespace prefixes
used within a document not match [known] URI schemes
... it will become common to sprinkle qnames throughout HTML
documents in the future
... if we do choose the q-attribute route, I prefer 'qcontent'
and leave 'href' alone
Ben: any other issues from IPTC discussion?
Mark: we need to settle our view
on reification
... we've revised the interpretation of ID so many times in
this design
... though I quite like the idea that ID refers to the
_statement_
... IPTC does want to be able to say _who_ made a statement,
_when_ the statement was made, how confident they are of the
statement, etc.
Ben: did custom attributes come up?
Mark: sort of, but I described it
as solved
... but there is a desire to treat, for example, role= as
identifying a property/value pair
... other things like media-type would benefit from such an
approach as well
... IPTC's requirement is for compactness
... they don't like 'property=x content=y' on every
element
... They've also defined elements ala 'dc:subject' with
attributes
... very RDF-like
... they also have the problem of Schema validation
Ben: it would be good to summarize our current thinking on these 3 issues in an email
Mark: we could use DanBri's
samples to help this discussion
... I've already noticed some possible improvements after
trying to write DanBri's examples
... IPTC has also promised some examples, including some very
big cases
... to help us understand the compactness issue
ACTION: Ben create a way to track progress on the 3 issues of qnames, reification, and custom attributes and elements [recorded in http://www.w3.org/2005/07/12-swbp-minutes.html#action01]
Ralph: would Misha be interested in formally joining the WG?
Mark: perhaps. They're certainly
willing to attend some TF telecons.
... but they probably won't want to attend every meeting
... there may be applications that want to do metadata
extraction by traversing an infoset via DOM