See also: IRC log
<DanC_lap> Welcome Jonathan, ashok
Those present record their thanks to DO for bringing us back to the big hole in the ground
SW: Possible addition to the agenda
for a telephone slot for a discussion of ARIA-related issues
... Proposed to put it into the tagSoupIntergration-54 slot
... May take quite a lot of time
TBL: We shouldn't land them with the
burden of long-term TAG issue resolution
... So we should review this in advance
SW: We could use last slot this afternoon for this. . .
NW: What kind of preparation?
TBL: Long-term planning, around issue of validation
SW: So, tsi-54 slot is confirmed for
call-in from Al Gilman and Michael Cooper on ARIA, possible one
other
... Other agenda issues?
HST: UAR-50 unlikely to take 90 mins
DC: Issue ScalableAccess-58 might benefit from some time
SW: Noted
SW: Want to get us moving on WebArch
v2 before next f2f
... Close some long-open issues, 34, 50, maybe 57
... Welcome other proposals for early closure candidates
TBL: I would like us to resurrect the Webarch 1.0 Errata doc't as a place we record things, for example 'resource' should be 'thing'
<DanC_lap> (note to timbl: the errata process starts with somebody sending mail to public-webarch-comments@w3.org )
<DanC_lap> (hmm... "n3 rules for how you follow your nose"... interesting.)
<DanC_lap> (pointer to the work TimBL attributes to David Booth?)
<Zakim> DanC_lap, you wanted to think out loud about the value of a one-time publication (webarch v2) vs a community/journal (TAGlines blog, W3C Q&A blog) and to offer that the TAG issue
<DanC_lap> . http://www.semantic-conference.com/
<DanC_lap> . http://www.w3.org/People/Connolly/#travl
<timbl_> I think the someday pile is important.
<timbl_> But it has to be so labelled.
<timbl_> "deferred" state
<dorchard> ping
DC: The issues list has value as it identifies for the community that certain issues are recognised, even if we don't know what to do about them
<timbl_> +1 for forward-facing radar
<DanC_lap> this meeting: self-describing web "all but done" and/or grows n3 rules in 6 months...
<DanC_lap> ... Bristol: hypertext tech blocks
<DanC_lap> ... this year: if people are talking about the TAG blog at influential ftf events
<DanC_lap> (urns and registries... what's the relevant ftf event for that? Norm, what about the XML UK event?)
<DanC_lap> (or some lifesci event, jar ?)
<DanC_lap> (as for XMLFunctions... I think that's a big one... feels like about a dozen N3 rules with 5 test cases each )
<DanC_lap> ("1st WD by december" tells me it's too big for a WD; a WD takes at most 3 months to produce)
<Norm> well, maybe it could be 3 mo if we started in earnest, I'm assuming there'll be a few months of ramp up time
<DanC_lap> (oh yeah... security... what happeend after the W3C mobile ajax workshop?)
DO: New topics we should be looking at, where there's a lot of innovation: social stuff, in particular
<DanC_lap> (re tagging... I'm fairly content that we're not talking about that... it uses webarch just fine. maybe we can use it as a hook to introduce our topics. re social, that's hugely important, but I tend to approach it more in my research work, though I'm speaking at KM Australia 2008 which is all about social stuff and tagging)
<DaveO> http://www.pearsonhighered.com/educator/academic/product/0,3110,013613517X,00.html
NM: Not sure that volume 2 is the for-sure right focus for WebArch -- possible _version_ 2
<DanC_lap> (hmm... I agree a lot of good stuff goes into findings, but the community review process is not all that smooth; www-tag sorta works for a crowd of a hundred or so, but I wonder if a different mechanism would increase the size of our audience substantially.)
<DaveO> (now should we be pushing JSON?)
<DanC_lap> (I push JSON a little in my research/dev blog(s), but I don't see it as all that influential on architecture, except in weird cases where JSONRPC has a different security model than XMLRPC)
NM: We can't _make_ anything a success, but we can and should look to providing the background/ammunition for groups who choose for their own reasons to move in the right direction -- RESTful WS are a success story in this regard
[TVR arrives]
<DanC_lap> (a few JSON items: http://dig.csail.mit.edu/breadcrumbs/node/96 http://dig.csail.mit.edu/breadcrumbs/node/99 )
NM: The new interactive media are a potential threat to WebArch: FLASH is qualitatively different from XHTML+SVG, and we need to look at that
AM: I come at the access to metadata
issue from a different angle
... but it's something I care about
... I'm also involved with OpenID in my day job, so would like to
hear what's going on there
... Also starting an IG on mapping relational data to RDF+OWL, so
there's possible interaction there
JR: Still trying to get up to speed
-- learning requirements from the TAG perspective is a goal for
me
... I'm on the hook to the HCLS IG for a document about URIs, and
although I
... am not happy with my current draft, I expect it's likely to
disagree with some TAG findings
HST: Hope we can talk about this under UAR-50. . .
JR: I think there are some standards missing which are holding up the SemWeb project
DO: What kind of standards?
JR: Ontologies -- foundational
stuff
... AWWSW for example, and bibliography and provenance -- lots of
duplication of effort in these areas
DO: Microformats guys did something like this, e.g. with vCard. . .
JR: Problem isn't technical, it's organisational
TVR: Not technical, same problem as AI has -- any success is no longer considered AI
<DanC_lap> (yup; when it works, it's no longer called AI. SemWeb has some of that. meanwhile, re "ignition", see http://esw.w3.org/topic/SeedApplications and http://esw.w3.org/topic/VocabularyMarket )
<DaveO> (I've also heard about Open Source Semantic Web, Drupal in particular)
JR: SemWeb only works if vocabs get shared -- I don't believe the 'precipitation' approach in which 10 different ontologies are built for the same domain is going to work very well, if at all
TVR: Hoping to make my finding on
issue WebApplicationState-60 as something useful for the Web
... Not by being proscriptive, but by collectiing and tabulative
current uses, detecting conflicts, and making best practice
recommendations
[Break until 1110]
[Resuming]
<Stuart> http://www.w3.org/2001/tag/2007/05/webarch2-outline.html (invisible on IE)
NW: I would start to fill this in with content, and only then address the 2nd edition vs. volume 2 question
NM: Both, I suspect
DC: Specific success criteria?
NM: Just as v. 1 put identification, interaction and ??? as the foundations of the Web, we add what we need for Linked Data and semantic reasoning.
DO: Wrt adding something to do with
Social Computing, maybe it's just applications of what we have
already
... but I'm not convinced. Consider stuff like Twitter for
example
<noah> Elaboration of above: Sample goal: Just as AWWW First Edition set in place the foundations of the Web itself, 2nd Edition will additionally provide equivalent conceptual foundations for linked data and semantic reasoning.
DC:Are you endorsing that as the right goal?
<DanC_lap> (I carefully avoided "the")
<noah> NM: No, not necessarily. I thought you asked for an example. That's a potentially goood goal?
DO: Or a tabulation I just added to my blog of all my 'activities' -- this on Flickr, that on Digg, etc.
<timbl_> Sounds like personal data integration.
<noah> DC: Can you give me something you would endorse?
DO: Maybe this could/should be mapped to RDF, so it could be merged, etc. . .
<noah> NM: Prefer not to now. I'd rather have the rest of the TAG iterate to the right high level goal. I'm not ready to say that I know what it should be just yet.
<DanC_lap> (re dave's points on I chaired a teleconference on data aggregation and syndication, hoping an XG would form; no joy)
TVR: Do we really need _architectural_ work to hook all these things (RDF, SemWeb, Web 2.0) together?
<Stuart> http://www.w3.org/2001/tag/2007/05/webarch2.oo3 I think
TVR: My preference would be just to
make sure that they all are based on the same architectural
foundations
... then integration should follow
TBL: New thing on the web: Oauth
SW: There's a theme here that maybe should be highlighted, which is activity
NM: Connects with Flash and
Silverlight
... Connects up with scheme/protocol issue, and with
selfDescribingWeb
<DanC_lap> (Dave, if you're interested in this "if you want to comment on my blog, you have to be a friend of friend..." stuff, see the DIG blog. http://dig.csail.mit.edu/breadcrumbs/taxonomy/term/4 )
<timbl_> BTW the interraction involved in HTTP is under the hood of the web as an informationspace
<Zakim> noah, you wanted to a) say self-desc is bigger than Formats and b) we need richer formats for Rich apps
TVR: Important to get level
right
... It would have been wrong 10 years ago to focus on shopping
carts
... Rather stateless vs. stateful
... leading to cookies
... So listing applications is the wrong level
DO: So better to talk about authentication, and managing the proliferation of identies
TVR: So focus on primitives in the
architectures
... e.g. that we need URIs for things, including identities
... If we could discover one more thing like like that, that would
be good
<Zakim> ht, you wanted to support Noah, I think
<DanC_lap> +1 move naming/URIs to the top of the outline
http://www.w3.org/QA/2007/10/the_impact_of_javascript_and_x.html
<DanC_lap> (Dave, if you're interested in this "if you want to comment on my blog, you have to be a friend of friend..." stuff, see the DIG blog. http://dig.csail.mit.edu/breadcrumbs/taxonomy/term/4 )
HST: Not sure when what a URI
'accesses' is determined by a rich interaction between Javascript
and XMLHTTPRequest
... not clear what is 'identified' by that URI
<DanC_lap> (hmm... http://www.w3.org/QA/2007/10/the_impact_of_javascript_and_x.html got 4 comments, all housekeeping. )
TVR: We may not have a meaning for URIs for application states therein, but that's an area we can work on
<Stuart> q/
TVR: I agree that the URI certainly doesn't identify some page
NM: But there are successes -- I can work with a Google map interaction for a while, and get a URI which reconstructs that for a 3rd party
TVR: and bookmark and Back
TBL: If you build everything you do on the basis of RDF, then by definition you get a URI for all aspects of the experience
<raman> on through biota -- thanks jar!
TVR: Time was when URIs were all the
same -- that is, there was no sense of a URI which worked on my
machine and not on yours
... but with the advent of history tokens tagged on the end of
URIs, that's no longer true
... For example with Dojo or GWT you can push tokens on the
interaction state, and sometimes the results are bookmarkable, but
the results are rarely emailable
... So more and more URIs are becoming dependent on
browser/platform environment, the evaluation environment
NM: Violates WebArch
<noah> From WebArch: "Since the scope of a URI is global, the resource identified by a URI does not depend on the context in which the URI appears (see also the section about indirect identification (§2.2.3))."
<noah> I think it's pretty clear that what Raman's been describing conflicts with that.
TVR: How we model this/modify WebArch is not clear
<raman> cd
HST: WebArch is just wrong on that: All file: URIs and some http: URIs, e.g. http://localhost/...
TBL: Those are edge cases
<timbl_> TimBL: The context-dependence of the file:// etc is a bug not a feature
NM: I'm not convinced that we need to relax the context-independence statement
TVR: Another example from Google --
URIs for identity -- Google calendar uses URIs for everything, you,
calendar, events, etc.
... The API will give you an Atom Feed
... Suppose your Calendar doesn't use https, but you wish they did
-- you construct https URIs, for the same events
[scribe is lost]
<noah> Let me clarify a bit. I said that I think we should try hard to keep the principle of Web Arch that the resource identified by a URI should not (except in oddball edge cases like file:) depend on context. I also said that with respect to local browser interaction models, "rich" interactions may (or may not) need from the user agent some richer history or navigation model than what a stack of context-independent URIs can supply.
<timbl_> TimBL: The fact that https: has a different scheme name is a bug too, though a bug we can't get out of.
<DanC_lap> ACTION: raman send email about growth of context dependence in URI interpretation [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action01]
<trackbot-ng> Created ACTION-105 - Send email about growth of context dependence in URI interpretation [on T.V. Raman - due 2008-03-04].
NW: It's a shame that https is a different scheme
HST: How do we connect the interest
manifest in this discussion back to WebArch 2.0
... I'm happy with the outline, although I'm terrified to open up
the Pandora's box of browser as platform
NM: Do we need to change the form in which we publish? Is continuing to publish findings obviously wrong?
<DanC_lap> +1 series. journal. blog.
<Zakim> timbl_, you wanted to suggest that we, like the first time, take the outline as a very roughg andchangeable framework, and be directed by where the pain is, wher the issues are and
<Zakim> jar, you wanted to ask if we know who audience is & what changes we want in their behavior
<DanC_lap> (the audience I had in mind for webarch v1 was: the typical W3C WG member, working on new web standards)
<timbl_> I would as co-chair point out that Jonathan and Ashok should feel free to use the benefit of their new eyes before they feel totally up to speed
JR: I would like to understand in each case how we are trying to influence people
<DanC_lap> (indeed; if "getting up to speed" means "making sure the TAG doesn't change", don't do that.)
NW: Worth a pass which adds a
paragraph, and connects up to issues list
... I will do that
trackbot, who do you know?
<DanC_lap> trackbot-ng, status
trackbot-ng, who do you know?
<scribe> ACTION: Norman make a pass over the WebArch 2.0 doc't which adds a paragraph, and connects up to issues list [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action03]
<trackbot-ng> Created ACTION-106 - Make a pass over the WebArch 2.0 doc't which adds a paragraph, and connects up to issues list [on Norman Walsh - due 2008-03-04].
[break]
i.e. ISSUE-41
daveo: summarizing past work
... q: how to version xml-based languages. started tactically,
wildcards in schemas, etc
... how to generalize beyond xml schema?
... how to generalize beyond xml?
... terminology of versioning. what is a version, language,
extension, consumer
... information conveyed
... after generalizing we came back to look at xml
... exposition was quite long.
... split into several documents
... another reorganization of docs: 1. terminology, 2.
compatibility strategies, 3. xml [jar not keeping up].
... hardest kind of compatibility: forwards. how to do compatible
evolution.
daveo: strategies document. tag had consensus partway through at last f2f. now working on book
<DanC_lap> (ah... a book... yes, this always felt more like a book, to me)
(see agenda for links)
daveo: SOA patterns book takes up the versioning theme
<DanC_lap> (do the book parts get noted in pacificspirit? I wonder about aggregation in the TAG blog again)
daveo: features in schema 1.1 can be used to support versioning
stuart: there are three documents in play here
daveo: compatibility doc is the one the tag should be able to finish up with
stuart: what we've done is to move the bar down through that document
daveo: was talking about strategies document
<DanC_lap> (pointer to the strategies document that dave nominates: http://www.w3.org/2001/tag/doc/versioning-compatibility-strategies
noah: what daveo wrote about schemas for schema wg is focused on explaining new features of schema 1.1
daveo: compatibility strategies document is 10 pages + 4 pages boilerplate. tractable.
daveo is going over the table of contents
daveo: soap, xslt specify things that
must be understood
... bulk of doc is on forward compatibility
danc: owl wg is doing work on versioning now. they have similar material - did a survey, picked a design
<DanC_lap> RIF WG on extensibility: http://www.w3.org/2005/rules/wg/wiki/Arch/Extensibility_Design_Choices
<DaveO> (Dan, I haven't posted the book stuff onto pacificspirit)
<DanC_lap> (ok)
danc: background docs were similar in
scope. ideally tag would synthesize their work & daveo's
... oops! I meant RIF, not OWL
danc: we should steal what rif has done, then ask them to review our results
daveo: a lot of this material isnot covered. could be disruptive.
timbl: might it be better to get daveo's document out first, then compare?
danc: rif should review what daveo
has already [this is not what jar recorded above. jar probably got
it wrong]
... it has become less clear that everyone wants the same thing
taking up section 2 of the compatibility strategies doc
<noah> I looked at http://www.w3.org/2001/tag/doc/versioning-strategies-20070917.html . Probably the old version, but Dave thinks not much text has changed.
<skw> http://www.w3.org/2001/tag/doc/versioning-compatibility-strategies-20071113.html
<Zakim> DanC_lap, you wanted to wonder about "None" in section 2... my mind wants an example
(everyone mulling over section 2, through but not including 2.1)
norm: docbook
stuart: html is going in the 'none' direction
<DanC_lap> (hmm... HTML WG as reviewers of this versioning stuff? )
danc: a point of contention
... we're discussing whether html is 'none' or forward
compatibility
timbl: forward compatibility
heated debate
daveo: had an argument with xxx about whether it makes sense to report an error and then continue processing (e.g. when extra arguments to a function are ignored)
noah: shades of gray between ignoring
something and having default processing rules. policy constrains
what you can do in future versions of language
... example: CSV -- cannot be versioned
... suggests splitting 'none'
stuart: into no stated strategy vs. no stated difference between versions [scribe is losing track a bit]
noah: there are no versions vs. there is no statement about versions (how they relate)
<DanC_lap> (I have made my peace with the "None" para, FWIW)
stuart: can we for each choice give one example of a language that makes that choice?
noah: hyperlink to definitions of terms (eg forward compatible), or give a brief informal definition?
daveo: html is an example of every category [yucks]
<timbl_> TimbL: Patterns. Peoploe understand and can re-use patterns. Re-use by analogy. Re-use with tweaks.
jar: how should this document be used - what behavior should it affect? (bears on question of whether to put examples into this exposition)
daveo: i wanted tag to advise on how to achieve compatibility; tag said no
<DanC_lap> +1 http://en.wikipedia.org/wiki/Pattern_language
daveo: so scope limited to setting up a structure for framing discussions. help wgs to talk about compatibility/versioning
<DanC_lap> ", an architect and author, coined the term pattern language. He used it to refer to common problems of civil and architectural design, from how cities should be laid out to where windows should be placed in a room. The idea was initially popularized in his book A Pattern Language."
<DanC_lap> +1 "design patterns for versioning" [and/or "xml versioning"]
noah: maybe call it 'design patterns for versioning'
daveo: maybe: keep the same
namespace, but break everything that uses it...
... while the language is under development
noah: be careful about assuming use cases. look at who the user community is etc
<DanC_lap> DO: I did write up 8 patterns
daveo: created the 8 design patterns,
including the forward compatibility pattern
... [see SOA patterns web page. link in agenda]
... tim, did this framework guide xml schemas and namespaces
timbl: for namespaces, they decided the question was out of scope
discussion of tim & dan's 1998 note on extensible languages
<skw> http://www.w3.org/TR/1998/NOTE-webarch-extlang-19980210
noah: question came up, would someone doing a new version of the language, change the namespace? (schema 1.0)
daveo: got thru 2/3 of a page in 42
minutes
... only 7 hours to go
jar: why uncomfortable reviewing: doesn't feel like fundamentals are solid. use of some formal tools would help me maybe
tbl, dc: we tried this and it didn't work
discussion of how daveo can find direction, based on feedback that tends to bloat the document and make work
doc was started in 2003...
daveo: it's worth spending 10 hours of together time to finish this
danc: how about if ashok and/or jonathan reviews what's left in detail?
(danc leading the process of harvesting commitments to review sections)
<DanC_lap> trackbot-ng, status
ACTION on daveo: Revised version of compatibility strategies document by next telecon (13 march)
<DanC_lap> ACTION: Dan review compatibility-strategies section 3 (soon) and 5 for May/Bristol [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action04]
<trackbot-ng> Created ACTION-107 - Review compatibility-strategies section 3 (soon) and 5 for May/Bristol [on Dan Connolly - due 2008-03-04].
<DanC_lap> ACTION: Ashok review compatibility-strategies section 2, 4 a week after DO signals review [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action05]
<trackbot-ng> Created ACTION-108 - Review compatibility-strategies section 2, 4 a week after DO signals review [on Ashok Malhotra - due 2008-03-04].
<DanC_lap> ACTION: T.V. review compatibility-strategies section 3, 4, 5 due 2008-04-10 [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action07]
<trackbot-ng> Created ACTION-109 - Review compatibility-strategies section 3, 4, 5 due 2008-04-10 [on T.V. Raman - due 2008-03-04].
<DanC_lap> ACTION: Norman review compatibility-strategies section 3, 4, 5 [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action09]
<trackbot-ng> Created ACTION-110 - Review compatibility-strategies section 3, 4, 5 [on Norman Walsh - due 2008-03-04].
<scribe> ACTION: David to revise version of compatibility strategies document by next telecon (13 march) [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action10]
<trackbot-ng> Created ACTION-111 - Revise version of compatibility strategies document by next telecon (13 march) [on David Orchard - due 2008-03-04].
<scribe> ACTION: Noah to review compatibility strategies section 2 due 2008-04-04 [recorded in http://www.w3.org/2008/02/26-tagmem-minutes.html#action11]
<trackbot-ng> Created ACTION-112 - Review compatibility strategies section 2 due 2008-04-04 [on Noah Mendelsohn - due 2008-03-04].
[convening again after break]
Stuart: introducing issue brought up
by ARIA. they want people to annotate scripts with info about
purpose, for accessibility reasons
... presenting email posted to www-tag
<skw> http://lists.w3.org/Archives/Public/www-tag/2008Feb/0087.html
(scribe got sidetracked by reading and listening)