TAG F2F, 3-5 March 2009

03 Mar 2009


See also: IRC log


TV_Raman, Larry_Masinter, John_Kemp, Tim_Berners-Lee, Henry_Thompson, Dan_Connolly, Jonathan_Rees, Ashok_Malhotra, Noah_Mendelsohn
Noah Mendelsohn
DanC, John Kemp



<DanC_lap> Scribe: DanC

<DanC_lap> ScribeNick: DanC_lap

NM reviews goals...

"Review F2F Goals

* Bring new TAG members "up to speed" on continuing work

* Make progress on high priority technical issues

* Establish TAG priorities for coming year - ensure issues list reflects actual priorities

* Refine TAG administrative procedures


TimBL: welcome to new tag members! I'm excited to get the burst of new momentum that comes with new members...
... and thanks Noah for chairing!

NM: by way of agenda review... not as much in the way of drafts to review in preparation for this meeting...
... thanks, ashok, for arranging the meeting facilities
... we'll start with active technical work on the 1st day before stepping back to look at overall priorities
... we'll talk about meeting schedule etc. this PM; if you can check your calendar during a break, that'll probably help

LMM: for this AM, I should defer input on priorities?

NM: well, I don't want to start 1st thing with wiping the slate and setting priorities, but this will be iterative

AM: conneg and redirections are related but scheduled separately...

NM: yes, they're related... we'll see... we don't yet have an issue on conneg
... minutes 19 Feb http://www.w3.org/2001/tag/2009/02/19-minutes ...

2009/02/23 21:03:34

NM: we'll look at that later


JK: John Kemp, working for Nokia ~6 years; prior to that, liberty aliance, OASIS Security Services Technical Committee; so my background in SOAP based Web Services, XML, ...
... web applications, a few start-ups doing software as a services; that goes back to ~1996
... I'm interested in the versioning and error handling stuff... and web application security

LMM: Larry Masinter I'm Adobe. Was doing metdata for video. Web Standards is now my full time job. In the 1980s I worked on the Common Lisp standard...
... I was at Xerox Parc when KR was done with ()s rather than <>s. Overlapped HT there.
... at Xerox we had experiments in networked information retrieval... then I went to a gopher conn... then I heard about WWW and downloaded a client...
... in the Common Lisp standards process, one of my main contributions was an issue form where you had to describe the problem independent of the solution and such.
... so I brought some of that experience to chairing the IETF URI WG and the HTTP WG.
... I was on the W3C advisory board and helped develop the TAG charter. I thought it was important to deal with conflicts between WGs and show leadership

TVR: Raman ... at Google... I care a lot about the Web and I'm concerned that the Web is being defined in terms of browsers too much; perhaps in reaction to being too far from browsers for a decade or so.

TBL: never mind history, where I am at now... when I find time for software, I work on systems where the Web and the Semantic Web are completely integrated...
... where systems are connected with other systems we haven't invented yet

<Zakim> timbl, you wanted to say Interested in arch coherence of the whole network of systems; the tech getting better not just older; very integrated sem web and web viewpoint; modularity

TimBL: I'm interested in the Web continuing to improve and involve and not fossilize

AM: Ashok Malhotra at Oracle, mostly doing Web Services...
... but I'm also looking at taking relational data and mapping it to the Semantic Web

HT: Henry Thompson; I'm half time W3C Team and half time U. Edinburgh. My TAG time comes from the U. Edinburgh time...
... some conflict with teaching duties this spring
... my background is in computational linguistics...
... a theme in my TAG work is to find ways to apply that background
... I have a number of paused TAG tasks; some because I'm stuck, some [for other reasons?]
... some of you know Harry Halpin, also at E. Edinburgh...
... Harry Halpin has now submitted his PhD thesis [applause]

<ht> scribenick: ht

DC: WebArch scales down as well as up, I've been learning about that with my new G1 phone
... Stuff about privacy, caching, web on hip connecting to the big Web
... HTML5 has a bit about offline apps
... The WebApps WG has published a WD on Widgets
... There's a Google OpenWeb advocacy group which I pointed ot the Widgets work
... Free Software background, purposely multi-platform
... I'm the official Team Contact for the TAG

NM: DC is the human archive of the TAG

<scribe> scribenick: DanC_lap

JAR: my background is in computer science, esp programming languages...
... was involved in capability security... and scheme standards

[much scheme/lisp experience in the TAG]

JAR: [something?] led me to the phrama industry, molecular biology, which led me to science commons...
... so I'm trying to help make the web better for science...
... URIs/naming, data integration, etc.
... I gather there's friction in using the web that we could do something about

NM: Noah Mendelsohn... was at IBM... operating systems... highly transparent distributed unix... XML and SOAP... Java at IBM when Java was an "emerging technology"...
... I enjoy the overlap between my personal interests, my employer's business interest and what the TAG does
... what Dan said about the smaller platforms... smartphones and such... I see a tipping point approaching

SKW: Stuart Williams at Hewlett Packard in Bristol, England. working in an HP labs group on Semantic Web... growing with the linked data stuff...
... naming and addressing is a focus of mine... from bits, bytes and nibbles... LAN mac addresses are 48bits and not [n] bytes...
... also some work on how you can introduce devices that didn't know each other; e.g. a phone and printer
... got involved in W3C... found myself elected to the TAG... found myself co-chairing; after the WebArch, took a break for a couple years, then another go at chairing... it's been a wonderful community to work with
... expect to continue to do related work

HTTP Content Negotiation (maybe track under ISSUE-57 or should we open a new one?)


Several recent email threads have raised questions about the proper use of content negotiation on the Web. ..."

<DanC_> (hmm.. URI for "view on this bug" vs "this bug"; no, doesn't sound like a case for conneg to me)

<raman> correct.

<noah> Yes, Stuart, thank you. The Martin Nally question is squarely within webApplicationState-60

<ht> HST: I've produced a review of http://www.ltg.ed.ac.uk/~ht/TAG_conneg_200903.html

(hmm.. copy should go to www-archive. here's hoping)

<DanC_> "The RDF response (modulo the lack of redirection) implies the resource is not an IR" huh?

<skw> Raman wrote a finding on Generic Resources which also bears on this issue... and I'm wondersing whether there was a TAG issues that that was written against.

<DanC_> (for reference: http://www.w3.org/2001/tag/doc/alternatives-discovery.html )

<DanC_> HT's notes on conneg for TAG meeting

HT: in sum, yes, I see issues in this conneg thread that should be re-opened or opened

LMM: background... in the '80s, the convention was naming, addressing, and routing... variant content types wasn't the norm...
... [Weiser?] had an idea of variant forms... Xerox work at the time asked these questions about "the gettysburg address" and such when tim visited us... I wonder if that's where Tim got the idea

TimBL: we could check the dates of my web architecture notes...

LMM: at the time, the idea was one HTTP transaction per click...
... there's some stuff in the HTTP spec that should be updated... just because the client knows 1000s of media types doesn't mean you should send 1000s of accept headers... "deprecate conneg" wasn't really what I meant
... I'm concerned that web architecture refers so directly to HTTP-specific mechanisms...
... questions of best practices should be approached with extreme caution...
... better to come up with descriptive findings "if you do X, then Y" rather than "you should X"

[chair seems to be writing on the whiteboard; scribe hasn't looked at it]

JAR: I think the advice from the TAG is pretty consistent and we could quickly address this...
... conneg came up "cool URIs for the semantic web", came up in "XRDs" which they have since abandoned in favor of [x?], came up in MH's question...
... this group's answer seems to be: we'd rather you didn't conneg in [which situation?]
... a test for conneg is: does it violate common expectations around a URI; does it lead a user to wrong [?] information. [?]

<timbl> Advice: Don't use conneg when it would mess up th eexpectations of normal web users

<jar> Test: If conneg might lead a use of the URI to the wrong result for some client, try to figure out a different way to do what you want to do.

<johnk> I agree with timbl's advice, however, a) what is the "wrong result" for a client that says it prefers RDF?

TVR: thanks for the nice summary, HT... on the generic resources finding... it was written using the old model of the web: one click, one http transaction...

<jar> see my email. The RDF might be viewable in a browser, so info provided by RDF should be same info as infor provided in HTML rep

TVR: then you got CGI... and conneg still works...
... but with active content, where the HTTP response is a program to run on the client, that turns the conneg situation upside down

<ht> HST thinks his TAG blog entry refers wrt Ramans point: http://www.w3.org/QA/2007/10/the_impact_of_javascript_and_x.html

<Zakim> johnk, you wanted to ask about TAG finding on authoritative metadata in this context

JK: the authoritative metadata finding has this notion of metadata in the container...
... is it relevant? does any of this [thread?] run counter to that finding?

<johnk> http://www.w3.org/2001/tag/doc/mime-respect-20060412#why

HT: the authoritative metadata finding tells what should be done; it's relevant in that it tells me that the punning examples are not at all compelling

JK: conneg is not just about the server saying what the server has; there are cases where the client wants only RDF and leaves the server in a position of [?]

<Zakim> timbl, you wanted to say conneg between metadata and a document it s is about is always wrong.

TimBL: Larry, yes, a lot of webarch is only implemented in HTTP. So while it's important to distinguish generic architecture for HTTP, it's also natural to talk about HTTP specifically
... I was a little disappointed that MH had to ask; I thought there was a community consensus that no, don't use conneg for [that]. So yes, perhaps we need to write it down.

<masinter> conneg definition was carefully hammered out in HTTP-WG, and some assumptions about it seem to be counter to the intent and (I hope) the written spec

TimBL: setting up the tabulator has been important in working out the practicalities of content negotiation...
... and for using RDF as a human-readable format

<timbl> 1) Much of web arch is only implemented in HTTP. 2) TAG giving advice is asked for and useful; 3) never use conneg between a document and metadata about that document.

<Zakim> DanC_lap, you wanted to suggest that HTTP is design as an embodyment of webarch and to suggest that the conneg story includes all 3 cases: server side, transparent, and client-side

LMM: I think it's useful to move beyond what is to belief/intent...
... getting that worked out in HTTP was tricky but worthwhile
... [image/rdf ?] is a big leap that I don't think we should be making

<DanC_> [darn; forgot to make my point about safety and POST and onload, which is the most important of the N things I q'd for]

AM: so this idea that conneg is only to be used for "equivalent" representations...
... am I hearing others say we should move away from that?

NM: I think [Martin?] is sympathetic to that position... that distinct URIs should be used, but he finds that when he builds it that way his users aren't happy

<johnk> is the only case where conneg is actually an arch issue in the link between data and metadata?

whiteboard experiment:


* AWWW discusses mainly the simple cases

* When conneg is used, how many resources are there?

* should "web" arch be independent of HTTP where possible?

* AJAX returns a program. conneg needs to be rethought for this?

(pertinent issues: generic resources & web app state)

* role of authoritative metadata & content type

* proposed test: "same information in different form?"


<DanC_> aha... I like that formulation of the test: "same information in different form?"

LMM: there's a section of the HTTP spec that's being revised... it's non-normative... about what conneg is for
... I have an action in the HTTP WG to work on that; I'm happy to get TAG input on that

TimBL: light-weight issue... yes... I'll help Larry

LMM: yes, same from the point of view of the speaker

. ACTION Larry: draft replacement for "how to use conneg" stuff in HTTP spec

<ht> LMM: and if you abuse that you'll confuse people

. ISSUE: capture "same information in different form?" test for conneg

NM: the ajax bullet seems still live... let's take that under web applications tate

<Zakim> ht, you wanted to say "clarify same info in different form" should be done

HT: another point to capture: ... everything that needs to be said about the relationship between relationships in the generic resources finding...

NM: that sounds like the "how many resources" bullet...

<Zakim> DanC_lap, you wanted to suggest re-using generic resource rather than making a new issue

NM: considering: to re-open the generic resources issue re the "how many resources" bullet


<trackbot> ISSUE-53 -- Generic resources -- CLOSED

<trackbot> http://www.w3.org/2001/tag/group/track/issues/53

RESOLUTION: to re-open Generic resources ISSUE-53

LMM opposed, TVR abstains [ I think]

<scribe> ACTION: Larry to draft replacement for "how to use conneg" stuff in HTTP spec [recorded in http://www.w3.org/2009/03/03-tagmem-irc]

<trackbot> Created ACTION-231 - Draft replacement for \"how to use conneg\" stuff in HTTP spec [on Larry Masinter - due 2009-03-10].

<ht> trackbot, status?

<ht> ACTION: ht to Follow-up to Hausenblas once there's a draft of HTTPbis which has advice on conneg [recorded in http://www.w3.org/2009/03/03-tagmem-irc]

<trackbot> Created ACTION-232 - Follow-up to Hausenblas once there's a draft of HTTPbis which has advice on conneg [on Henry S. Thompson - due 2009-03-10].


HTML and "TAG Soup" Integration (ISSUE-54)


* Review history, successes, and challenges with respect to TAG's efforts ..."

JK: mixing XML languages with HTML... is that the goal?

<Zakim> ht, you wanted to review the founding statement

HT: "Is the indefinite persistence of 'tag soup' HTML consistent with a sound architecture for the Web?" -- http://www.w3.org/2001/tag/group/track/issues/54

<Zakim> DanC_lap, you wanted to say that space for unstructured discusion helps

DanC: space for unstructured discussion here helps balance social dynamics in the HTML WG

LMM: is this an HTML issue or a webarch issue?
... is extensibility, versioning, error handling... are these HTML problems or webarch problems?

<masinter> well, they're all webarch, but are they restricted only to HTML

HT: I noted "Is the indefinite persistence of 'tag soup' HTML consistent with a sound architecture for the Web?"; but note also " If so, what changes, if any, to fundamental Web technologies are necessary to integrate 'tag soup' with SGML-valid HTML and well-formed XML?"

LMM: yes, this is an architectural issue, but is it wider than HTML? [?]

HT: it's wider because it's the thin end of a long wedge...
... extensibility mechanisms in HTML are likely to be picked up elsewhere

<Stuart> Larry's "no" above was in response to an aside question "Stuart wonders whether when speaking of HTML Larry is also including XHTML?"

HT: what I heard in the ARIA discussion was "we don't need extensibility because extension happens rarely"; applying that generally is at the very least a general architectural issue

<jar> (noodling on what raman's saying: html = 'shell' for the OS=web ?)

TVR: [missed some] a model was: how can we make a web where lots of languages can play at the end of one link?
... in the 1990's, we thought mixed namespaces, DOM, events bubbling, was the way things would work...
... more recently, others say no, [this other thing] is how it works
... does this mean we need to re-design SVG, ...

<Zakim> DanC_lap, you wanted to comment on XML and HTML, esp RSS, SVG, RDFa

DanC: on the one hand, HTML is just one among many content types in web architecture, but on the other hand the web _is_ HTML to 3 or 4 significant digits...
... and [more... even though larry repeated it...]

<Zakim> timbl, you wanted to suggest creative commons as a nice case in point, whcih suggest that there si no lace where we can draw a clean lin between html and xhtml

TimBL: some say "HTML is its own thing, not XML"; on the other hand, RDFa is designed in the XHTML context ...
... then a discussion was sparked by Creative Commons suggesting use of RDFa in HTML, regardless of whether it's XHML or HTML...

<masinter> is this issue about "namespaces" really?

TimBL: so attempts to keep XHTML and HTML separate have broken down

<Zakim> raman, you wanted to add that we need to remember that like namespace extensibility at its time, tagsoup today is also an experiment. Some would say that the experiment is not

TVR: we've taken the namespace-based architecture as orthodoxy for a while, but that was an experiment as much as the tag soup approach...
... for all we know, either could fail in the 4 year timeframe...
... if you look in the 10 year timeframe, we should acknowledge that experiments will fail... and we should look for ways that they can live together [close to that, anyway]

LMM: the name of this issue misled me...
... I think maybe it's re-considering namespace based architecture
... a story: somebody came to me a the W3C TPAC and said "we need a registry; how do we make one?" I tried to make a joke about "there's this way of doing decentralized naming..." but they didn't get the joke
... namespaces were introduced as a way of decentralized naming... they were rejected for perhaps technical/usability reasons... [scribe lost train of thought here]
... one position is "within this context, we don't need distribution; we can manage the chaos by getting all the browser developers in the room/group" ...

<raman> union of namespaces was proposed by people like Tantek multiple times

LMM: we probably need union namespaces

<DanC_> (reminds me of prospero... union links are a great idea; I wonder if they can ever get widely deployed in filesystems. )

<Zakim> ht, you wanted to mention the sniffing draft


<trackbot> ISSUE-24 -- Can a specification include rules for overriding HTTPcontent type parameters? -- OPEN

<trackbot> http://www.w3.org/2001/tag/group/track/issues/24

<masinter> jar: seen something like this before -- Common Lisp package system


HT: perhaps we would re-open the authoritative metadata issue for that

DanC: yes, issue-24 was re-opened for that

NM notes it's in an "unscheduled" pile in the agenda

NM: I think decentralized naming using URIs is architecturally important, but I've never seen a user-friendly syntax for it...
... looking at Java packages, people can see the use case for DNS-based naming when they borrow code from elsewhere
... but there are screw cases with bug fixes across domains and such

<DanC_> (+1 look at both sides)

NM: I think both sides are making important points and we should take both seriously

<Zakim> noah, you wanted to point to lack of agreement on need for extensibility

NM: [something about convenient syntaxes being less self-describing ...]

<masinter> I don't think we can abandon namespaces, but without namespaces, may need registries or some other clear extensibility mechanism

TimBL: something is either self-describing or it's not...
... I think we can solve the problem with manageable cost for all the relevant parties
... HTML parsers are already huge; a little more code to handle namespaces is a negligible cost...
... [an example elsewhere; scribe missed; help?]
... I think yes, it's a goal to get the creative commons feature working on HTML 5

<masinter> +1

TimBL: by whatever technical or social means necessary

<Zakim> ht2, you wanted to query state of media-typed based NS defaulting

<Zakim> timbl, you wanted to suggest etchnically that html5 adopt ns for new browsers.

<johnk> agree with timbl, masinter that a good concrete goal is to get RDFa working in HTML5

HT: this idea about default namespaces based on media types has been discussed in "we should..." mode, though I don't know that anybody's actively working on it
... I'd like to establish a reward for cleaner markup in _this_ life...
... it was at best naive to expect XHTML would dominate; but we don't have to give up on the idea that XHTML has real benefit.

TVR: absolutely

HT: I concur with the idea that tim has presented: let's remove the step function in the reward of cleaning up HTML markup. [?]

<johnk> is there a link to that work?

<masinter> not sure "media type based namespace declaration" is the right formulation

<timbl> You can addd xmlns for cc and get the benefit of having cc markup without having to put quotes around attribute values everywhere for example.

LMM: not sure "media type based namespace declaration" is the right formulation... but perhaps formulate it in terms of mapping rules between contexts
... i.e. how to interpret one as the other

<Zakim> masinter, you wanted to separate costs to readers, cost to authors

<Zakim> noah, you wanted to discuss wrapup

NM: I made some notes on the board [scribe wishes for a photo in www-archive] but let's not go into those

LMM: let's keeping working toward a deliverable... finding/note no matter
... esp. on conversion between namespaced and non-namespaced forms

TVR: I think the TAG has a critical role in bringing 2 sides of the community together

<Zakim> DanC_lap, you wanted to note HTML WG agenda for this week

close ACTION-226

<trackbot> ACTION-226 Report at March on tagSoup progress since TPAC closed

<noah> DC: Chris Wilson has an action HTML WG action to write up distributed extensibility

<masinter> trackbot, action-226?

<trackbot> Sorry, masinter, I don't understand 'trackbot, action-226?'. Please refer to http://www.w3.org/2005/06/tracker/irc for help

<noah> ACTION-226?

<trackbot> ACTION-226 -- Dan Connolly to report at March on tagSoup progress since TPAC -- due 2009-02-24 -- CLOSED

<trackbot> http://www.w3.org/2001/tag/group/track/actions/226

<noah> scribenick: noah

TBL: Henry mentioned talk I gave at AC meeting suggesting both sides should make some concessions to come together on this. I can review it.
... It met with some criticism that it's asking too much of both sides, but my opinion hasn't changed.

<Zakim> johnk, you wanted to ask about timbl's idea of addressing the CC use-case in HTML5

TBL: Henri Sivonen pointed out that current HTML browsers don't actually populate DOM namespace properties.

<masinter> if it is parsed as an HTML document

<Zakim> noah, you wanted to discuss use cases

JK: so, you think discussing the CC use case is important.

TBL: yes

<timbl> Hmmm. yes sorry that is in fact not totally written up ... it has bullet point in it

<DanC_lap> Cleaning up the Web

<DanC_lap> (crud; I marked the essay world-readable, with permission, but I didn't link it from http://www.w3.org/2008/10/TPAC/TPDay-Agenda.html )

<timbl> Those are notes from the talk, but should go with the slides

<DanC_lap> you didn't use slides when you presented it at TPAC 2008

<jar> danc: The namespace skeptics are in the minority in the marketplace.

DC: Dominant browser ships namespace-based extensibility. Similar in spirit to XML namespaces, but not exactly XML.

TBL: Does it use xmlns?

DC: not remembering

<masinter> Chris Wilson from MS will report on this in HTML WG

<DanC_lap> HTML WG ACTION-97

<DanC_lap> Following SVG-in-HTML thread, propose decentralized extensibility strategy for HTML5

HT: What I want us to do is: "produce a document which specifies a mechanism for bridging the gap between namespaced and non-namespaced forms of languages conveying the same things"

<DanC_lap> break for lunch, resuming at 1:15pm

<DanC_lap> scribe: John Kemp

<DanC_lap> scribenick: johnk

URNsAndRegistries-50 ISSUE-50

HT: introduces issue, noting languishing of this issue


ht: intends that this doc is the one to be blessed
... origins of this issue were concerns about uri scheme and namespace proliferation
... OASIS eg. uses URNs for XML namespaces
... thus no straightforward way to dereference
... notes URI scheme examples such as XRI
... wrote a doc to defend the proposition that http: was sufficient for naming
... concern was that original doc would not actually help those who needed it most
... there is an opinion that everything relevant to be said is said already in the relevant RFC already
... Dirk & Nadia attempt to document the story a la AWWW
... example: boss says we want names, Dirk says let's use URNs, Nadia says new scheme!
... biggest problem is the taxonomy of requirements: e.g. what is persistence?
... (describes document structure)
... not worked on this doc recently
... but wanted to introduce this to new members and give us a memory jog

NM: (asks the room what we think)
... not sure the description reveals the issues
... gives HT the chair for this session

LMM: describes link on problems URIs *won't* solve

<masinter> problems URIs Don't Solve

LMM: people confuse name assignment with service level agreement about name resolution
... when you buy domain name, you get a service guarantee
... (discusses control and monetization in name assignment)
... biggest piece of puzzle from Dirk & Nadia is issue of control

biggest piece of puzzle _missing_ is issue of control

LMM: new URI scheme, or URNs may provide more benefit than cost for some

DanC: URI space owned by all, not "some"
... issue of namespace collision

timbl: (challenges)

LMM: use uuids
... (see XMP uuids)
... they weren't locators

<ht> No-one ever expects the network effect

DanC: http is good for domain + path hierarchy

<DanC_lap> ... but uuids are outside that case

LMM: look at lifecycle of content and its movements

HT: introduction of Dirk & Nadia needs to say that interest is in naming things with a Web context, not other naming

is widget naming "for the Web"?

<jar> danc: anyone smart enough to use large random numbers doesn't need to go to school at our school

DanC: people want to recreate a naming hierarchy, is the issue

HT: previous document set out to sell HTTP, Dirk & Nadia provides instead a cost/benefit analysis
... lookup + hierarch will meet most requirements, and HTTP URI already provides that
... intent of doc is to show you how to sharpen your reqs, and if you use HTTP URIs, how to meet them

LMM: distinction between a film and the (actors) in it
... identification problems are different
... identifying media means identifying the content container
... identifying concepts described in the content is different

HT: doc not written to address those problems

JAR: assumes you have your own theory of naming

LMM: no one has a well-defined theory of naming

JAR: there are some theories that work

(in practice)

<Zakim> noah, you wanted to ask if we've documented costs

NM: have we documented or understood the _costs_
... to what extent do you plan to dig into the costs of alternatives?

<jar> in reply to Larry who said xxxx is not well-defined: well-defined and adequate are not the same thing

HT: moved costs to 'spare parts' section

<jar> what I mean to say is that the theory of what is named is orthogonal to naming system mechanics. can solve the 2 independently

HT: large analysis of the tradeoffs - how you get confidence in certain guarantees, by contract or otherwise
... that level of analysis did not belong in the finding

NM: namespace pollution is only part of the issue
... ... other reason is association of HTTP URI with widely-deployed dereference mechanism
... two parts of this story could be told fairly simply
... your decisions affect other people (if you take a name, no-one else can)
... wide deployment of existing schemes may be useful to you

<Zakim> masinter, you wanted to ask: who needs this finding? What W3C work depends on an answer to this question?

LMM: prioritizing - whose work depends on us answering this?

HT: we were asked by W3C members

NM: will discuss explicit priorities tomorrow

TVR: (echoes LMM)

JAR: (thinks a contribution can be made here)
... ... particularly when trust is lacking, naming is an issue

HT: D&N does address that somewhat (by noting checksums in URIs)

JAR: we should talk about priorities, issue is important in SemWeb

<masinter> *IF* we could resolve this effectively *THEN* there might be value

TBL: rate of non-HTTP ways has remained steady

LMM: (mentions TDB, DURI schemes in this context)

HT: next decision points comes with a next draft

LMM: [trust, authority, control, monetisation] all go together

Review of ongoing TAG projects

NM: one time only (we will not repeat this in future meetings)
... we had 28 open issues at time of writing agenda

<masinter> ... as motivating factors for why people want new schees


NM: believes that some issues are open because we think there is something to be done here, but in practice it is not clear what we should do

<DanC_lap> (not just to remind ourselves that it's open, but as a marker in the community that yes, you're not the only one with this problem)

NM: proposes we close issues which fall below some mark
... and sort the rest appropriately
... (wants to get others involved in managing the issues well)
... (profers the issues list at http://www.w3.org/2001/tag/2009/03/03-agenda#issuetable)
... proposal to schedule two sessions with break
... first session, divide into groups
... new TAG members circulate between groups
... (shows tracker)
... (introduces tracker functions)
... nowhere does it say in tracker item where are we?

DanC: (expresses enthusiasm to move forward)

<timbl> http://www.w3.org/2001/tag/group/track/issues/open

<noah> deliverables are we expecting to produce

<noah> http://www.w3.org/2001/tag/group/track/agenda

NM: describes how he creates the agenda
... some day it would be nice if the summary of agenda input were more to the point
... shows a template for describing issues
... (describes template)

TVR: we need to figure out the meaning of the criteria

LMM: what other activities depend on the work?

NM: let's take ISSUE-50
... and work a real example

HT: where do you propose to put the info (from the template)

NM: show tracker fields are very fixed
... put this information in the description field

JAR + LMM: use Notes instead?

NM: 'notes' falls to the bottom when read

TBL: limit each of the things in the template to one line?

LMM: does priority belong in description?

(working ISSUE-50)

NM: priority: 'medium'

(discussion about whether this is right)

TBL: which are the issues that we're doing while Noah is chair?

NM: concurs
... (describes how he will handle issues)

AM: can we add a field?

NM: tables the question
... and all other 'meta' questions
... trying to capture where we are accurately

LMM: what is the priority, and how certain of it are you?

NM: OK to say 'don't know'

DanC: no way to find priority in our (past) records

NM: use your collective consciousness to work it out
... in many cases, you know

LMM: what do you think the priority of each issue was?
... vs. what do you think the ongoing priority _should be_?

AM: questions how we do this for very long-running issues

NM: (back to ISSUE-50)
... edits tracker item - http://www.w3.org/2001/tag/group/track/issues/50

<DanC_lap> ("current deliverables: ...namingSchemes" is redundant w.r.t. actions; I suppose that's sorta by design.)

(group discusses communities affected by ISSUE-50)

HT: 'raised by' now means 'principally responsible'

NM: external commitments on ISSUE-50?
... goal is to do all issues by 16:00
... (proves, by PIE, that not all administrivia is bad)

(group breaks into two to divide issues list)

(no more minutes for now)

<jar> re issue-24: http://lists.w3.org/Archives/Public/www-tag/2002Jun/0116.html

<timbl> http://www.w3.org/TR/WICD/

<DanC_lap> issue-30: http://www.w3.org/2001/tag/issues.html#binaryXML-30

<trackbot> ISSUE-30 Standardize a "binary XML" format? notes added

<DanC_lap> issue-30: http://www.w3.org/TR/exi-best-practices/

<trackbot> ISSUE-30 Standardize a "binary XML" format? notes added

(back to minutes)

NM: summarizes what happened wrt the tracker issues
... would like to appoint a shepherd for each active issue to help prepare related agenda items

JAR: could it be done via ACTION assignment?

NM: would propose we do add an action for each one which is not in shape

each one == each issue

LMM: issues not open don't need shepherds
... some issues should be moved from 'open' to 'raised' (or 'closed')
... ones marked as background don't need shepherds

NM: "au contraire"
... let's go over the issues and decide how to resolve

Issue status: whenToUseGet-7, etc.

DanC chairing this session

NM: go till 16:30
... (moves to open issues)



HT: HTML WG work on ping attributes is believed to be moribund

DC: (disagrees)

HT: Seems unlikely we need to do anything about this
... Revive if discussion is raised again in HTML

(discussion over who should shepherd)

LMM: could we close it?

<timbl> NM, 2009-03-03

DC: does anyone volunteer to declare victory?

HT: do we want to give that message?

NM: if we leave it open, I'l monitor this

<timbl> NM, date +"%Y-%m-%d"

LMM: worthwhile to discuss what it means to close an issue
... communicate to community what it means when we close an issue

HT: sending any such message in this case would be inflammatory

NM: history about how we use the 'open' designation
... community has expectations based on that designation
... do other TAG members agree with Larry?

LMM: if we close a number of issues at once we're sending a different message
... not saying anything specific about any particular issue

DC: in this specific issue (ISSUE-7) feel compelled by Henry's argument

TBL: There are issues with 'ping', and then close it
... see the whenToUseGET finding
... notes other issues (beyond using GET)

NM: Is there consensus over text I'm typing?

(seems not)

DC: Is this proposal to close?

NM: just to agree text for current description?

DC: (asks group for thoughts on decision to close)

HT: (some dissent)

NM: don't want to change the criteria for closing

JK: would abstain

NM: if we believe we should keep an eye on some issue, we keep it open
... if no obvious reason to come back to the issue, then close it
... I would rather look over definitions of open and closed over time
... that is not the goal for this session

TBL: am I allowed to ask about how we use tracker?

DC: polls whether to close ISSUE-7

NM: describes 3 possible actions mostly missed by the scribe

DC: is this text (ISSUE-7 description) what you would like on record as resolution for this issue?

NM: keep it open, is text OK?




JAR: kept open in case RFC 3205 was too restrictive

LMM: BCPs not normative

NM: what is the TAG's role here?
... appoint shepherd

DC: not worth the effort needed to close it
... mark pending review


come back to ISSUE-20 tomorrow


HT: what does description text mean?

DC: AWWW says something about this

<scribe> ACTION: Larry to report back from IETF/HTML liason meeting in March regarding MIME type override [recorded in http://www.w3.org/2009/03/03-tagmem-irc]

<trackbot> Created ACTION-233 - Report back from IETF/HTML liason meeting in March regarding MIME type override [on Larry Masinter - due 2009-03-11].

DC: connection to tagSoup unclear, happy to delete from text
... shepherd?

LMM: (nominates himself)

NM: will discuss role of 'shepherd' later


DC: rank high is surprising

HT: open action actively pursued

NM: high rank is related to a big piece of work we will do this year

HT: medium is OK for me

NM: (marks it as medium)
... (adds more context)

LMM: issue is misnamed

DC: proposal?

LMM: "Use of IRIs in W3C Specifications" as new title

NM: do we mean only W3C specs?

TBL: "when should IRIs be used"?

LMM: don't want to talk about IRIs in email e.g.

NM: what have we actually been doing?

<ht> LMM and DC please note http://www.w3.org/TR/leiri/ wrt your ACTION

TBL: everywhere you use URI you should use IRI

(that's what we've been saying)

DC: this is an involved discussion
... suggest we move on

<timbl> s/verywhere you use URI you should use IRI/We were saying "verywhere you use URI you should use IRI"/

HT: no title proposal has yet attracted support

old title: Should W3C specifications start promoting IRIs?

NM returns as chair

NM: this work is important, even if unpleasant
... proposes we put the exercise down for now, perhaps return Thursday

<DanC_lap> (I wonder about a shotgun approach to sheperds...)

NM: designation approach for naming shepherds

DC seems to agree ^

NM: goal is come out with a more refined view
... any concerns about this?

JAR: if you want to get through the list come up with an offline procedure to follow

NM: will do that and see how it goes

<Zakim> DanC_lap, you wanted to propose to close httpSubstrate-16 and to suggest merging fragmentInXML-28 with abstractComponentRefs-37 and to note 12 May 2004 decision that puts 28 in

AM: fragment in XML brought up by WSDL group
... this is water under the bridge

<DanC_lap> PROPOSED: to close issue-28 on the grounds that WSDL 2.0 is a REC

NM: will entertain a proposal to close ISSUE-28

LMM: mark pending review
... will look at it

<DanC_lap> (+1 pending review an LMM to look at it)

<DanC_lap> RESOLVED: to close issue-28 on the grounds that WSDL 2.0 is a REC

Administration: TAG Operations, Scheduling Future Meetings

NM: http://www.w3.org/2001/tag/coordination/TAGGuide.html
... (describes reasons for writing the document)
... are we willing to agree that this describes how we will work?
... reads text about good scribing practices to help the chair
... also updated document to note that draft minutes should include plaintext

LMM: sent feedback that document combines procedural things with TAG "philosophy"

NM: would you like to rewrite this?
... Will think about your feedback
... intend this to be a hitchhikers' guide to the TAG

LMM: procedural issues may have no longevity, philosophical ones may have more
... perhaps should be separated for that reason?
... if you're asking whether this is a good starting point, then agree

NM: (would like to have practices that he can "enforce")
... do you buy the goal?
... do you have any objections?

JK: what actually though are the long term goals of this document?

(essentially agreeing with Larry about the impression of mixing two separate, perhaps both important but perhaps better separate subjects)

<DanC_lap> (I have a bias against standing items.)

AM: (raises procedural issue I didn't really catch)

TBL: concerned that this doc is not public

NM: I want this doc to be for us, not for public feedback
... but if we decide that our public commitment is obvious elsewhere would be OK with this being public
... asks whether he should schedule additional work on this document or whether it works roughly

(group agrees to work with what is written)

NM: next TAG meeting scheduling?
... would like to do another f2f meeting around summertime
... any sympathy for Boston area meeting for early June?

HT: have scheduling difficulties generally

NM: proposes 2/3/4 June

16/17/18 June?

27/28/29 May in Boston?

or 17/18/19 June

TVR: Is this just about picking dates?
... can we do this in email?

two concrete proposals: i) 27-29 May, ii) 17-19 June

23-25 June?

PROPOSED: tentative 23-25th June meeting

RESOLUTION: to check this against the previous proposals

NM: adjourns meeting

Summary of Action Items

[NEW] ACTION: ht to Follow-up to Hausenblas once there's a draft of HTTPbis which has advice on conneg [recorded in http://www.w3.org/2009/03/03-tagmem-irc]
[NEW] ACTION: Larry to draft replacement for "how to use conneg" stuff in HTTP spec [recorded in http://www.w3.org/2009/03/03-tagmem-irc]
[NEW] ACTION: Larry to report back from IETF/HTML liason meeting in March regarding MIME type override [recorded in http://www.w3.org/2009/03/03-tagmem-irc]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.134 (CVS log)
$Date: 2009/04/02 20:42:59 $