See also: IRC log
Regrets from Vincent, TimBL, Henry for 8 November
Ed Rice volunteers to chair the 8 November telcon
Vincent will prepare the agenda
<DanC> I'm OK to scribe 8 Nov.
<DanC> regrets 15 Nov due to XML 2005
<DanC> ok, color me at risk
Regrets from Vincent for 15 November, Ed is at risk, Dan Connolly at risk
HST is at XML 2005, hopes to make the call
Noah will chair on 15 November
Regrets from Norm for 15 November
VQ: Nominate RF to scribe on 15 November
<noah_sfo> FWIW: we had the compoundDocs. stuff scheduled for last week, and left it off this week in part on the assumption that I would not be here today. I note that Tim is unavail. next week. Maybe or maybe not it's worth trying to slip it in during the first hour or so today while I'm around. If so, give me 3 mins notice to find materials. Either way is fine with me.
<noah_sfo> OK, I'll go find cd stuff
Norm: June f2f logistics are being sorted out, looks good
Added new agenda item on mixedUIXMLNamespace-33
<DanC> 18 and 25 Oct minutes OK by me
RESOLUTION: 18 October minutes approved
RESOLUTION: 25 October minutes approved
<DanC> (18 and 25 oct minutes edited to remove DRAFT)
HST: These are not the slides that I'll speak to at the meeting, rather what goes to the attendees in advance
RESOLUTION: Approve http://www.w3.org/2005/11/tag-summary.html for publication once the missing link is filled in
<scribe> ACTION: HST to produce a draft of slides for TAG slot at AC meeting in time for telcon on 22 November [recorded in http://www.w3.org/2005/11/01-tagmem-minutes.html#action01]
Attendees at AC meeting: Vincent, Noah, Henry, Tim, Dan
[Note that Dave Orchard will be phone-only for December f2f]
NM: Took an action at f2f to review CDF by-reference Requirements and CDF Framework and WICDD Profiles
NM: Sent
http://lists.w3.org/Archives/Public/www-tag/2005Oct/0040.html my review
to www-tag,
which was not specifically based on mixedUIXMLNamespace-33, but
covers most of that stuff
... CDF WG has prepared two things: 1 Requirements doc, 1
Profile
... The latter has been split subsequent to first
publication
... They distinguish between by-reference and by-inclusion
compounding
ER: Why?
NM: Not clear -- they thought by-reference was going to be easier, but empirically it seems to me that that's not at all clear
TBL: Inclusion means the
mixed-namespace-thing
... which is our
http://www.w3.org/2001/tag/issues.html#mixedUIXMLNamespace-33
... Reference means e.g. using an Object, which doesn't raise
all the hard questions
NM: Makes sense
... Core of our concern is the general semantics of
mixed-namespace documents
<timbl> "Self Describing Documents on the Web"
NM: To what extent is a random mixed-NS document self-describing?
<DanC> "WICD, or Web Integration Compound Document, is a specific embodiment of CDF using XHTML, SVG, and CSS."
NM: CDF is focused on UI-oriented stuff, for a bounded set of vocabs, focussing on HTML, SVG, etc.
<timbl> Noah: SDDOTW is something we the TAG should get into moreo and more
NM: So e.g. what happens when you
click on a bit of real-estate rendered from a nested bit, does
it propagate out, etc.
... I wish they'd layered this much more, separating out the
general question from this more focussed one
... I've had private communication from a WG member which is
sympathetic to this point
... Sounds also like the WG is not actually working much from
the Req'ts doc. . .
TBL: Thanks Noah
... Counterarg't to the generalisation you argue for --- there
are some things you can't do in XML until you know the
semantics
... I think mixedUI case is one of these
... Consider RDF, the semantics of mixing is clear because it's
been designed in
... Similarly wrt the mixedUI case, because it has an
underlying coherence, they can talk about e.g. what happens to
mouse clicks
... But in general, w/o that semantic info, you can only say
something very shallow/weak about composition
... Similarly, this feeds over into our discussion of
versioning, languages, etc.
... E.g. when a language has styling semantics, some versioning
stories make sense, but don't necessarily generalize
<Zakim> DanC, you wanted to question advice to generalize, unless we're confident there's implementation experience with something more general
DanC: Don't want to set requirements they can't meet, and the general problem is too hard to hand to them
<timbl> It is difficult to go far talking about versioning without more semantics than raw XML. Hypertext and Semantic web are example of language domains in which there are enough semantics to do more.
DanC: Examples I'm aware of (OpenDoc, Andrew) are not getting used
<Zakim> noah_sfo, you wanted to respond on generalizing, and to point out the earlier systems have separately layered compound document semantics from UI
NM: What about OLE, I'm using it
<Zakim> ht, you wanted to disagree with conclusion
HST: Tim said what I was going to say
<Zakim> DanC2, you wanted to ask whether the CD requirements doc expresses a position on xlinkScope-23 and to ask what became of VQ's investigation of the CDF WG's media type issues
DanC: did they take a position on the XLink scope question?
NM: Not as far as I remember
NM: In reply to TBL et al.'s point
[missed some of Noah's comments]
NM: OLE has an abstract notion of
hierarchical (stream) story, modelled as a 'baby' FAT file
system
... If you crack open e.g. a Word document, you find such a
thing, with the analogues of QNames connecting things up
<Zakim> dorchard, you wanted to ask about composition of transforms, specificially xinclude + xslt + ...
NM: CDF people could/should do something similar, was my point, w/o boiling the ocean
DO: Compositionality of transforms point? XInclude, XSLT, etc.. . .
NM: Not that kind of transforms, rather scale/rotate/displace
DO: Anything about the XMLProc sort of transforms?
NM: Maybe, would need to check
DC: Vincent, what was answer about media type issue?
VQ: Haven't asked yet. . .
<Zakim> Roy, you wanted to ask if anyone remembers why this is a TAG issue and whether CDF is aware of it
RF: Why was this issue split out from a larger issue, what are we expecting from this group, are they aware we're expecting it?
TBL: We're not expecting something from them specifically, but we should be keeping an eye on what they're doing
RF: So I'm happy that Noah checked it, but what's it to do with us?
VQ: It does overlap with mixedUIXMLNamespace-33
RF: So why didn't we just close the issue as soon as the CDF WG was formed?
NM: Well, so much discussion about self-describing at the E'burgh f2f, I took this on in that spirit
RF: OK, that's clearer
<Zakim> ht, you wanted to say they're the wrong people for the job
HST: I don't think the CDF is the right group of people to ask for a story about general semantic composition. OLE isn't a general story, it's still a UI/presentation focussed story
<Roy> I suspect that the reason we split the issues is specifically so that the WG could be formed to address this, not the *general* issue which has a different number on our list
NM: Well, I am still concerned that they can't do their job if they don't layer it better. But I hear the input I'm getting, so we're not going to feed that in to the WG as such
<Zakim> timbl, you wanted to say yes to Noah, the XML functions paper really concludes that that top-down model of XML semantics is very important. But maybe the TAG should say it rather
NM: But there are other things in my email that might be useful, so can we point them at it 'officially' w/o endorsing the layering point
TBL: Talking about semantic
composition is important, but the CDF WG shouldn't be asked to
do it generally -- I still hope the TAG will tackle that
problem, under the heading of mixedNamespaceMeaning-13
... We'll pend our mixedUIXMLNamespace-33 issue until CDF have gone a bit
further, and then have another look
NM: But what about those on the WG who want to go in the generalizing direction?
TBL: I think we should encourage them to focus on the UI-specific stuff
NM: Well, that will disappoint some people
<dorchard> I'm in favour of non-generalization for CDF.
DC: Straw poll: a) more discussion; b) withdraw mixedUIXMLNamespace-33; c) close it on basis of CDR WG existing; d) pend until CDF make more concrete progress
NM: What do you prefer?
<timbl> I support (d)
<dorchard> dave: option d). I'd rather talk about other things.
DC: (d) plus encourage them
<Roy> my pref is (d)
HST: (d)
<Norm> NDW: (d)
<Ed> D for me as wel
<Vincent> (d)
<noah_sfo> Fine with me.
<noah_sfo> I could live with other options, but (d) is just fine.
RESOLUTION: Pend mixedUIXMLNamespace-33 until we see significant progress in a public draft from the CDF folk
<scribe> ACTION: NM to follow-up to CDF that his email was not endorsed in detail by TAG [recorded in http://www.w3.org/2005/11/01-tagmem-minutes.html#action02]
<noah_sfo> ACTION: Noah to send note to CDF clarifying that earlier note was just a draft, that TAG has not to provide formal input on CDF Reqs at this point, and hinting at concerns about generalizing [recorded in http://www.w3.org/2005/11/01-tagmem-minutes.html#action03]
VQ: Mark Nottingham thanked us for our input, said they'd welcome more input in the next few weeks
HST: Will circulate a worked example as soon as possible, hope by end of week
VQ: Right, so what do we do to help them?
HST: We need to talk about this in any case, if we miss their deadline so be it
VQ: They have a f2f next week, at
which point they will have a better sense of their
deadline
... So we'll return to this next week
<DanC> (for reference, our last discussion of ns8 http://www.w3.org/2005/10/11-tagmem-minutes.html#item06 )
<DanC> usps
<DanC> e.g. http://www.w3.org/2000/10/swap/pim/usps#MailPiece
DC: The above URI tries to do
both the RDF property thing and the document fragment
thing
... But it's also a class name, and you can use it in RDF
TBL: Is it a convention, or is it trying to really identify two different things in two contexts?
DC: I don't what to say I'm using
the URI for two different things. . .
... Creative use of the HTML mime type
TBL: Loath to go there, because
it rules out writing RDF statements about anchors
... Back in June at MIT, we were working towards a position
<DanC> re rdf statements about anchors, http://esw.w3.org/topic/HashSlashDuality
TBL: That when you use a fragment in a UI context which refers to e.g. a class, you fall back/coerce to a presentation you can see
DC: We have to hide one (the
anchor) or the other (the class)
... I just want to make this OK
HST: Tim, could you clarify your fall-back story wrt the redirect advice we gave for the namespace URI
<DanC> (the TOC of the minutes should show fragmentInXML-28 as much as issue 8, please)
TBL: Depends on whether what it
gets from the redirect is RDF or HTML
... That in turn depends in part on what the agent is that's
asking
DC: So we all think this is OK? I don't hear anyone saying it isn't. . .
TBL: It's wrong because it uses the same URI for 2 different things We're exploring the options
DO: Is this related to the
abstractComponentRefs-37 as well
... and the question the WSDL WG asked about whether they could
refer to components or sub-trees
... and we said if they were careful with media types, then
yes
<timbl> The "fallback" philosophy (which I don't necessarily support) would be that the object is *really* a class. In this case, as the user agent can't grok RDF, and so can't get the full info on the class in the ideal from, there is a fall-back, a sort of type coersion, and an HTML document is delivered and the user directed to a paragraph about that class. There is some information loss, but this happens when you change content type. For example, if we gave a re
<timbl> A content negotiation with some degradation.
TBL: The danger is that the URI gets bookmarked and used purely as a pointer into the [HTML] document
DC: Is it reasonable to think of changing the HTML spec. to use the profile attribute on the document element do make the determination?
HST: How would this help?
DC: Problem was that if we put
this in the RDF spec, the browser never sees it
... So we put it in the HTML spec so it has to
TBL: Reverse engineering like that is difficult
DC: OK, thanks for airing this, content to wait until Norm gets back to writing
VQ: So we pend this
VQ: Once again no time for abstractComponentRefs-37 -- are we ready to talk about this in detail yet?
<DanC> (well, we did touch on abstractComponentRefs-37 )
DC: I had an action, which I satisfied last week, so I'm OK with not talking about it again right away
HST: DO has said we're done, because we replied to WSDL WG
<scribe> ACTION: DO to re-send pointer to basis for this statement about WSDL WG having gotten their answer wrt xxx-37 [recorded in http://www.w3.org/2005/11/01-tagmem-minutes.html#action04]
VQ: Then maybe we can close that issue
DC: My question is whether we'll say anything more to the WSDL WG - is Roy going to say something more?
<Roy> me? not that I can think of at the moment
<DanC> about #frags(like_this) in WSDL. are you likely to say anything more than your existing XPointer comments, Roy?
HST: I thought the question of XPointer syntax was part of another issue, where we were waiting on Roy
<DanC> ok.
<DanC> issue goodURIPractice-NN or some such.
<Roy> I have two findings to write, yes
<DanC> ok, we'll stay tuned. sooner is better.