hi ben
<benadida> Hi Dan
<benadida> alright, I am a Zakim newbie still (Ralph usually helps me), so what do we need to do to start logging and all? ah thank yhou
its all in hand
we're logging
who are we expecting today?
<benadida> Steven
<benadida> (yes, we miss you Ralph)
<benadida> Steven is on his way in
steven: last week I keynoted at News Summit in
Amsterdam
... 'set cat amongst the pigeons' (I'm told), in that their doc formats
hardwire metadata vocabs, so need revision for new properties.
... very rigid
... nice thing about rdf/a is that you can add new metadata at will
... i'll be talking to the iptc agm next week in london, on same topic
... also talked at xtech conference
[...]
I had some agenda fodder in http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2005May/0047.html
1) grddl/xslt'ing rdf/a
2) rdf links
ben: thx for rdfs... would be good to get
jeremy to take a look
... steven too?
<benadida> ACTION: Steven and Ben to take a look at the RDFS DanBri put together [recorded in http://www.w3.org/2005/05/31-swbp-minutes.html#action01]
danbri: some fiddly decisions... what namespace uri, what syntax to use, etc (see comments in the rdfs)
steven: doesn't really fit in the html doc itself
ben: could make it a tf note
steven: part of the reasoning behind this
is.... much as with HTML 4 profile, we defined things separately
... since detail could change, and be new version
... change to the thing at the namespace
... mechanism is extensible
... after CR, we won't add unqualified properties
... need to use a new ns
... alternate relationship... can specify conditions
... by language, media type, ...
danbri: would be nice if we could derrive my doc from your doc via xslt
mark: yup
ben: is this separate doc just an rdfization of the htm
danbri: that's the intent
(I tried not to get too creative... but there are some points that need review)
danbri: i wouldn't be averse to pushing this out as a note
steve: if you want to take our stuff and reprsenet it as rdf, the html would be happy
ben: could be a perfect candidate for a tf note
steven: am suprised there's no standard notion of 'document', so you had to invent your own
danbri: i was trying to avoid ratholes
danbir: we can't appeal to Docuemnt=xml Document
mark: stylesheets etc...
... also there are other relationships
... js etc
... also <style src= attribs, or <script ...
therefore when we define Document, we should be clear it isn't just XHTML2 or XML doc
ben: moving on
... i'll look into this being a tf note
... Integration with GRDDL
... steven, mark... could you consider adding something to the namespace
document for xhtml2 to point to the XSLT that turns XHTML2 RDF/A into
RDF/XML?
(conventions per the GRDDL note)
http://www.w3.org/TeamSubmission/2005/SUBM-grddl-20050516/ has details
http://www.w3.org/TeamSubmission/2005/SUBM-grddl-20050516/#ns-bind
danbri: your plans for the xhtml2 ns doc?
steven: whatever tim said!
some xhtml doc there
ben: right now you don't plan to have a schema
doc there?
...
mark: looking at various schema languages, so complicated
steven: we'd have to look at how you do it in other schema language
(some discussion of use of some modules)
mark: eg you could write an atom schema that importted bits of xhtml2
(detail of how GRDDL interacts with schema modularity)
mark: if i made an xhtml2rdf transform, what's the difference between that an grddl?
ben: same thing, really
mark: we'll have to make that xslt anyway
... I had an xslt early on... but we've changed the architecture since
... less complex nesting now
... only outstanding thing now is qname processing
... my proposal was to write functions that echo the forthcoming xpath2
functions which manage qname-based strings
ben: lets bring the xslt people back
together
... this is the right point
danbri: +1
mark: i don't have much time, but ... i'm
inclined to have a go, as a final test
... edge cases only became obvious after writing the xslt last time
<scribe> ACTION: mark talk with jeremy on public-rdf-in-xhtml-tf re xslt parser for xhtml2/rdf notation [recorded in http://www.w3.org/2005/05/31-swbp-minutes.html#action02]
danbri: getting that xslt will improve testing, (a) as we'll have a focal point for defining the mapping (b) many more people can explore test cases, queries etc.
ben: bnodes...
mark: i just posted my summary of the
situation
... steven and i aren't completely against the 2 attribute solution
... but the problem (URIs/IRIs) really lies in RDF's world, not in the
markup
... and you have this quirky situation of trying to id a node without using
global IDs
... how do you hide that? an rdf/uri problem
... not that we ignore it
... but adding more attribs takes away from some of the simplicity
... what we have now is very simple, based on earlier html conventions
... adding new attribs to encode different ways of identifying makes things
more complex
... the moment you name something with an ID, you get annoying situation that
you can't talk about it yourself
... the outside world can talk about it
... but ID="x" prevents me from making global statements
... doesn't work
<Steven> http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2005May/0049.html
mark: so i've come back to xpointer soln
... not sure if the objections to it were on basis of what it's really
arguing
... comments from dan and jjc
jjc - doesn't like it aesthetically
scribe: i do like it, as takes things into
realm of uris
... dan's msg is abuot it creating confusion
mark: to be clear, that the xpointer be
dereferenced
... we don't store that uri as a triple
... the serializer shoudl go off and find the node with the id=x, and then
genrate a local bnode for it
... cloaked request to the serializer to hide this node
... generate a uri that's different to the one you'd normally generate
... you get a local name for that node
(is this a skolem function, or do i have my terminology wrong?)
<Zakim> danbri, you wanted to suggest the issue will be encountered in terms of attribute content, not attribute name
<link about="#some(a)" rel="foaf:mbox" href="dan@example.com" />
<link about="#some(b)" rel="foaf:mbox" href="libby@example.net" />
versus
<link about="#some(a)" rel="foaf:mbox" href="dan@example.com" />
<link about="foo#some(b)" rel="foaf:mbox" href="libby@example.net" />
<benadida> ACTION: Dan to summarize discussions about bnode on mailing list (fallback to Ben) [recorded in http://www.w3.org/2005/05/31-swbp-minutes.html#action03]
<benadida> Dan, can you do the generation of meetings notes?
rssagent, draft minutes
ah thx
<Steven> and now: