SV_MEETING_TITLE -- 13 Sep 2007

<Stuart> :-)

Convene

<DanC> SKW: one recent change to agenda; any other mods?

<DanC> minutes 6 Sep

<DanC> RESOLVED to approve minutes 6 Sep (2007/09/06 18:55:56)

Next Telcon

<DanC> PROPOSED: to meet again 27 Sep, Rhys to scribe

<ht> Regrets for 2007-09-27

<DanC> RESOLVED: to meet 27 Sep (scribe to be confirmed)

September F2F Preparation

<DanC> http://www.w3.org/2001/tag/2007/09/17-agenda Revision: 1.14 $ of $Date: 2007/09/11 09:54:43

<DanC> SKW: Monday part of the meeting will stop at 5pm sharp if we're to take up the offer regarding a social event

<DanC> HT notes weather merits warm clothing

<DanC> SKW: monday morning scribe?

<DanC> HT: OK, I'll scribe Monday 17 Sep AM

<DanC> Rhys: regrets 27 Sep [?]

<ht> For wind and weather forecasts, I recommend http://www.xcweather.co.uk/

<ht> There are weather stations for both Southampton and St. Catherine's Point

Issue UrnsAndRegistries-50 (Tracker ISSUE-50)

<DanC> . ACTION-33 Henry: revise URNsAndRegistries-50 finding in response to F2F discussion

<DanC> HT: my work on that got preempted; sorry.

<DanC> HT: yes, I saw Chime's comments.

<DanC> HT: the http://esw.w3.org/topic/HttpUrisAreExpensive was news to me; thanks

<Noah_Bangalore> I'm afraid I really need to go. FYI, I made some limited progress on the plane on the way over in reading both the terminology and strategies parts of Dave's versioning drafts. Whether I'll be able to wrap them in an email before the F2F is not clear, but I'll certainly try.

Issue contentTypeOverride-24 (ISSUE-24)

<Noah_Bangalore> See you Monday.

<Stuart> Thanks noah...

<Noah_Bangalore> Actually I'm interested in this one, I'll stay a bit.

<DanC> DC: our position on contentTypeOverride-24 is: if it says text/plain, it's text/plain

<ht> ScribeNick: HT

NM: By "ask for an image" do you mena in the accept header?

DC: No, rather where the GET comes from, i.e. the markup around the link
... I can't remember whether the HTML 5 spec says anything about accept header
... I then thought about the overlap with HTTP
... and drafted an internet draft containing the sniffing rules
... which was enough to get Roy Fielding to join the HTML WG
... and some substantive discussion is now happening
... Ian Hickson says that having spent two years trying to do the right thing, and losing
... He believes that any browser that doesn't sniff will lose market share
... Fielding disagrees
... Hickson has in the past suggested the TAG would have more credibility if they reopened the finding and added orange cones, that is, "in practice" exceptions

<Zakim> Noah_Bangalore, you wanted to noodle a bit on accept headers vs. <img> tags

SW: I've always understood that the mime type as delivered is a statement of server intent, and that doesn't _entirely_ determine what client use must be

NM: What would a new story be, if the HTML WG gets their (currently specced) way? If they add an accept header, that makes some sense

but if _not_, I'm really worried -- it means you're interpreting the outcome of a protocol on the basis of something _outside_ the protocol

SW: Should we actually open this up?

<Noah_Bangalore> I'll try to clarify a bit what Henry scribed (you can fold this in when editing minutes of you like).

<timbl_> I wouldlike to push back on the HTML WG

TVR: I'm not sure this is a good idea, on the grounds that we've already opened up the HTML issue, and we don't yet know what the result will be

<Zakim> DanC, you wanted to concur with TVR about the risk here

<Noah_Bangalore> What I was trying to say was: it was asserted that the semantics HTML 5 wants is something like "if the link was from an <img> tag, then make different assumptions about whether the returned representation is an image, regardless of Content-Type".

TVR: Until we know that the HTML WG will succeed, I'm concerned about putting even more of WebArch at risk

<Noah_Bangalore> What bothers me about that is that not only is that different from HTTP as specified today, you can't even specify it in terms of information that's visible at the HTTP level.

<timbl_> Who at microsoft is the person who has to make that decision?

<Rhys> I agree with Raman

DC: This is an issue which makes me worry about the viabililty of the HTML WG -- Ian Hickson's position is that we can't do anything that the major browsers won't come on board with. Roy Fielding's view is that that means abdicating responsiblity, and just standardise other folks bugs

DO: I agree that this is not in the whole community's interest
... We had a less than completely successful attempt to do better in the Web Services area, I don't think we should roll over here as well

TBL: It's true that if we can't get the big vendors to change their minds, we're in a mess. I think the right answer is for the TAG to convince them to try to make their browser help users do better
... I'm prepared to put some efforts into make this happen
... For example, browsers should warn users when show cleaned-up source on Show Source or when they have to sniff
... Important to get this right _now_ -- this morning we were talking about video, and for a number of practical reasons, content negotiation between the two major approaches is going to be _very_ important
... and this is going to happen again and again, so we _should_ take up the challenge and try to persuade Firefox and IE and so on

<Zakim> DanC, you wanted to report a bit of hope for a mozilla build with an option to make bogus mime types visible to the user, and meanwhile, a fairly serious proposal to reduce the

<DanC> http://www.w3.org/Protocols/HTTP/1.1/rfc2616bis/issues/#i81

TVR: Sniffing is a slippery slope to disaster

DC: HTTP WG is thinking about restarting at IETF
... fixing bugs, issues list, no WG yet but close
... Larry Masinter has filed an issue to deprecate content negotiation
... Julian Reshke said he'd like a configuration option which said "show me the true mime type"
... That seemed like a hopeful sign

<timbl_> ... application/xml;dammit

DC: There has also been discussion [where?] of replacing or down-grading or ??? the Content-Type: header, e.g. application/xmlDamnIt

<timbl_> I don't see at all how that will solve the problem that ISPs don't allow folks to control the MIME heders

SW: So reopening means strengthening our arguments, or considering changing our position?

<Noah_Bangalore> +1 to reopen

DC: Could go either way

<Noah_Bangalore> or at least a strong concur

<DanC> DC, Tim, Dorchard, Noah...

SW: In favour of reopening: 4

<Rhys> Rhys doesn't have a strong view

SW: Abstain: HST, SW, RL
... Opposed: TVR

<timbl_> +1

TBL: Do we have to reopen the issue to discuss it?

TVR: I'd like to discuss it, but not change it

<Zakim> ht, you wanted to say enforcement

<DanC> (if we just want to re-assert our position, I don't think we need to re-open it.)

<Noah_Bangalore> I'm discouraged. For years we've asked ourselves whether the time is right to tell the story of the new parts of the web, like SemWeb, in an AWWW vol 2. Now it feels like we're deciding how much of V1 to withdraw. Sigh.

HT: I think it's always in order to discuss how to best promote TAG findings

DC: Not opening it says we're not listening

<Noah_Bangalore> Not that I'm against openning the issue, just discouraged that we need to.

TVR: Opening it says we were wrong

<Noah_Bangalore> I also don't think that openning an issue signals a change. It signals a careful recheck, I think.

DO: I guess opening it is sensible because it gives us a clean way to discuss, have actions, etc.
... There's a precedent in what we did with xxx-7

<Noah_Bangalore> I do have to go now. See you in Southampton. Good night!

DO: So if we reopen and say "We're doing this because there's new information, and we want to track that and interact appropriately"

<DanC> +1 re-open (I think the economics of the issue merits re-opening it)

SW: Asking again -- should the TAG reopen the issue

TVR: What does opening it mean?

HST: That it stays open until we close it or abandon it

<timbl_> TBL: yes

SW: In favour: DC, DO, TBL

Abstain: HST, TVR, RL

DO: If we're not going to open this, we shouldn't talk about it

HST: What about opening a new issue on "How do we deal with the fact that the HTML WG is heading down a road that is incompatible with our finding on respectMediaType-???

TVR: I like that

DC: I could live with it, but I think it's odd

DO: Same here -- it seems like a heavy burden on our process

SW: Proposal to reopen fails -- only three in favour

<DanC> ScribeNick: DanC

<ht> HST observes that there _was_ a majority in favour of talking about this matter. . .

SKW: proposal to re-open issue 24 didn't carry; other proposals are welcome. enough for today...

Tech-Plenary: TAG Proposal of Discussion Topic

<timbl_> ________________________________________________________

DO: [something] was accepted... 45 min...
... panel... with Q&A... <= 5 ppl... maybe 4...

<Stuart> ok it's written as: The Importance URI based Extensibility

DO: on the panel should be advocates of short strings as used in microformats, somebody from HTML 5, etc.

<Stuart> but the that seems to be different words for what has also been called distributed extensibility

DO: not sure if my role is just recruiting panelists, or MC, or participant, or what...

<Rhys> I think it's fine for the MC also to be a participant in the panel.

TVR: goal of the panel?

TVR: goal of the panel? this is clearly a long-running discussion.

DO: re-inforce our message in support of decentralized language evolution

DanC: goal is at least getting more of the community up to speed, if not achieving a whole lot of novel technical progress on the panel itself

TVR: ok, outreach makes sense

SKW makes suggestion to mitigate the risk that presentations would use all the time

((diversion back to ftf logistics; SKW notes Monday PM 5pm stop time))

((yes, pick up XMLversioning at 3:30pm Monday))

((same time window on Tues harder to predict, but agenda calls for Tag Soup))

Issue XMLVersioning-41 (ISSUE-41)

<ht> ScribeNick: ht

DC: I'm still interested in the thread on forward/backward compatibility definitions

<DanC> ACTION-4 on Dan Connolly to Review definitions of partial understanding, backward compatible, and forward compatible [DONE]

DC: Mark DeGraaw [sp?] has recently raised a real use case here, coming from HL7

DO: I've tried to add definitions, based on information, but that did not find favor

<DanC> DanC: my review (Fri, 24 Aug 2007 16:57:37 -0500) said "I don't find this appealing; looks like an open research problem". DO said "let's not formalize it that deeply" which seems ok, perhaps, to me and Noah... as long as it doesn't come up in the practical examples in the scenarios part

DO: Since I can't take it any further, I suggested dropping it
... I'm also interested in Mark DeGraaw's work, since there's a clear indication there of the value that success here would provide

<DanC> DanC: Marc de Graauw is working on a formalism; I found it somewhat interesting as an academic exercise, but much moreso now that he's pointed out that it's grounded in a real-world scenario: HL7

SW: HST, any input?

HST: I know of no solutions from the Computational Linguistics side

<DanC> ACTION-27 on Dan Connolly to ask Mimasa and Mark Birbeck about feasability of using substitution groups in XHTML modularization, cc public-xml-versioning [DONE]

DC: I asked, I didn't get a satisfactory answer
... Subst Groups are designed for distributed extensibility

<scribe> scribenick: danc

HT: I'm up to speed here... I read the modularization [of XHTML] spec...

(which see ACTION-15 )

<ht> http://www.w3.org/2001/tag/2007/09/xhtml-modularisation-thoughts.html

HT: indeed, substitution groups are a great mechanism for distributed extensibility; I explain how...
... a couple problems, 1 minor and 1 major...
... putting things in multiple substitution groups isn't allowed in XSD-the-REC; it is in XSD-work-in-progress...

<timbl_> "clean subsetting"?

HT: but they [XHTML] have another goal, that substitution groups don't do: "clean subsetting"
... substitution groups are bottom-up, which is why they're great for decentralized extensiblitiy...
... but XHTML modularization is also top-down; I dunno how to do that with substitution groups
... I hope somebody finds a work-around

DC: this "clean subsetting" ... not sure I understand the motivation...

HT: you need it to build XHTML 1.1 out of modules

[DanC isn't able to grok that right away]

DO: this is an interesting point on the design of XSD 1.1
... seems like XHTML is an important use case for XSD

[my understanding is that XHTML is _not_ widely regarded as an important use case for XSD]

(ACTION-15 is done to my satisfaction. ACTION-15 on Henry S. Thompson to Review XHTML Modularization )

HT: I'd be more interested in addressing that requirement in XSD 1.1 if I could see a clear design.

DC: DO, seen my msg about SMIL?

Subject: lots of SMIL namespaces, revisited [XMLVersioning-41 / ISSUE-41]

Date: Fri, 31 Aug 2007 12:52:24 -0500

DO: yes; saw that; been working on something related...

Issue abbreviatedURIs-56 (ISSUE-56)

ACTION-34 on Stuart Williams to Look at the difference between QNAME in XML and SPARQL [DONE]

SKW: SPARQL uses "abbrevited names" which are similar to QNames, but not quite the same. [summarizing http://lists.w3.org/Archives/Public/www-tag/2007Sep/0036 ]

DanC: recent RDFa designs seem to not use CURIEs in the href attribute, but only in new attributes

some investigation into current status of curie, RDFa drafts, inconclusive...

<ht> scribenick: ht

<DanC> Issue contentTypeOverride-24 (ISSUE-24)

back to contentTypeOverride-24 (ISSUE-24)

DC: TBL asked the question: Who makes decisions about this sort of thing for the vendors?
... We're planning a f2f of the HTML WG for the Tech Plenary week
... and we are hoping that there will be real technical representation from all the vendors

<timbl_> "sensitive mode"

TBL: I don't know what a non-obtrusive way to improve the media-type problem

<Rhys> 'view problems' in the same sense as 'view source'?

<timbl_> TBL: I think showing only a clean XML version of veiw sourcfe for copy/paste is non-destrictive

DC: The best suggestion I've seen was to allow the "This is my content, show me problems rather than fixing me silently"

<timbl_> Firefox > Tools > Error Console

DC: Roy Fielding is taking the HTML WG discussion seriously, but I don't know how far Ian Hickson has yet succeeded in changing his mind
... Who knows how to make a private build of Firefox? None of the people who do on the WG have stepped up so far. . .

DCandTVR: Boris Zbarsky seems like the best bet. . .

<timbl_> http://mozillalinks.org/wp/2006/10/do-you-know-boris-zbarsky/

web.mit.edu/bzbarsky/www/

- DRAFT -

SV_MEETING_TITLE

13 Sep 2007

Attendees

Contents