Admin

1 August call cancelled

8 August, DanC to Chair, regrets from Norm Walsh, Vincent Quint, Tim Berners-Lee

TimBL missing for all of August

VQ missing for last three meetings in August

Minutes of last telcon (http://www.w3.org/2001/tag/2006/07/18-minutes.html)

Approved.

<noah> Seeing that we will have spotty attendance in August, with some people gone the whole month, do we need to plan for any progress on the Oct. 3 F2F logistics during the month of Aug., or can it wait for Sept.?

Today's agenda: http://www.w3.org/2001/tag/2006/07/25-agenda.html'

VQ: Late addition: URNsAndRegistries-50

NM: Request slot at the end to comment on metadata-in-URI
... Planning for Oct. f2f -- leave it until September?

<DanC> you can get your tickets now, no? "face-to-face meeting, 4-5 Oct 2006, Vancouver, BC, Canada, hosted by BEA" -- http://www.w3.org/2001/tag/

DO: Logistics page will be available soon [now available] -- meeting at the Opus Hotel confirmed
... We will get special rate as long as we've got six people staying

<dorchard> Meeting at the Opus Hotel with 6 rooms blocked.

DC: DDR workshop report is out - - should we have agenda item to discuss?

<DanC> . http://www.w3.org/2005/MWI/DDWG/workshop2006/report.html

VQ: Add an agenda item on that if time allows

VQ: I've produced a first draft of our required quarterly report
... will circulate shortly
... and send if no comments

New media types

VQ: First draft out from NM: http://lists.w3.org/Archives/Public/www-tag/2006Jul/0027.html

NM: Background: There are WGs writing WDs which intend to reference common-but-not-registered media types (e.g. audio/wav)
... They're worried that they may get pulled up for this at review time, and so they asked the AB

NM: The AB asked the TAG if there was a technical aspect to this

NM: TAG discussed this last week -- our conclusion was that in some cases it's really not reasonable to expect WGs to take up the responsibility for dealing with unregistered types
... We got some pushback from publishing a draft to that effect
... Saying basically: "You use it, you have to pay for it"
... I.e. do the work to get it registered

TBL: Bjorn H. says that the process is easy, but I'm not sure it's easy enough to do this

NM: I thought the history of some of this was arose from commercial initiatives, which have passed in to common use

DC: If there are reasons to be impure, that's on a case-by-case basis, the general policy is "Don't do that"

<DanC> audio/vnd.wave;codec=1

<DanC> per

<DanC> http://www.iana.org/assignments/wave-avi-codec-registry

<DanC> and

<DanC> http://www.ietf.org/rfc/rfc2361.txt June 1998

NM: but it wouldn't be appropriate for e.g. me to register them

DC: The reason audio/wav isn't registered is that there's not enough information there

<raman> DanC I found your comments/reactions offensive. I'm trying to bring a sense to the TAG as to how WGs life is difficult -- sneering at things saying they should look at IANA isn't a good response

NM: This is one of the examples which Ken Laskey included in his original question

TV: I think you risk alienating the WG if you push that, it's just the kind of thing which makes WGs not think well of the TAG

NM: I thought to register a type you have to have some authoritative knowledge to register a media type
... Are you really saying that there's no minimum bar for what you need to know?

DC: Not in practice -- no reason to want to have something in wide usage and not register it

TBL: NM is asking whether what amount to 3rd parties can/should register

NM: Yes

DC: Well, just do it, and the people who should have done it can step in and take over

<noah> From my email:

<noah> * Accordingly, workgroups should in general arrange for registration of new media types that they may create, and should make reasonable efforts to promote the proper registration of other formats on which their Recommendations depend.

<Norm> I don't know where you said it, timbl_, but I read it. The XSL and XML Query WGs are putting the registrations in the specs.

TBL: My original goal was that all W3C specs should be just like an IETF registration, meeting all the requirements for them
... [scribe missed something ending: ... "it breaks"]

NM: Right, that's fine for things you're creating within the w3c process which include a new media type -- you better do all the necessary things to register it
... But the cases at hand are different -- we have cases where a WD is not defining a new media type, but needing to refer to an existing one.

<noah> Also from my email:

<noah> * Workgroups preparing Recommendations should in general make reasonable efforts to avoid dependencies on media types or other data formats that are not properly registered with the appropriate registration authority. In the case of MIME media types, that authority is IANA.

TVR: Right -- and we shouldn't tell such a WG that they have to take the load of cleaning up behind the people who should have done this

NM: [reads from http://lists.w3.org/Archives/Public/www-tag/2006Jul/0027.html]
... which concludes with the most controversial bit, which describes the basis for the exception to the "should" above

DC: I haven't heard any examples motivating the last bit -- just leave it out

TBL: Suppose the W3C had a list of the media types used in w3c specs but not registered with IETF

DC: This is a good use of whose time?

TBL: Well, we could then refer to this on IETF liaison calls

NM: Remember the original question was to the AB regarding the Process

DC: We should not give them this out

TVR: I don't believe we should say that
... Someone has to clean up the mess, but using "touched it first" as the way to tell who's the one to do it is a mistake

DC: We don't need to do anything to allow people to get excused a requirement at the final review

TVR: WGs see such prohibitions as real prohibitions, they don't think in terms of breaking them and making the case for that

TBL: An example?

TVR: Not a media type, but [xxx scribe missed]

TBL: I'm still looking for an media type example. If there are lots of these out there, that's v. different from the situation if there are only a handful

<DanC> (re timbl's "suppose we had a list..." comment, there is such a list in http://www.w3.org/2002/06/registering-mediatype.html ; Martin used to maintain it. I don't think it's worth much of my time. I think PLH spends some time on it.)

HST: I don't see a technical issue here, beyond "The TAG believes in follow-your-nose, unregistered media types break that, registering media types is a Good Thing"

<DanC> noah, we can say "follow your nose is important" without suggesting any sort of MUST/MAY re process.

HST: NM, is that message a draft of a message to the AB; or for the AB to resend as if it were theirs; or as a TAG statement to the public

NM: Great question! Today's discussion makes me realize that I wasn't sufficiently conscious of the target audience for the note. It probably should have been the AB, but I wound up writing as if it was aimed directly to www-tag and hence directly to the Web community. I think the response we actually craft should be formally addressed to the AB and should be written accordingly.

<DanC> perhaps reply to the AB by citing http://www.w3.org/2001/tag/2004/0430-mime

DC: What about http://www.w3.org/2001/tag/2004/0430-mime
... Could we cite that to the AB

<noah> W3C Working Groups engaged in defining a format follow How to Register a Media Type with IANA [IANAREG] to register an Internet Media Type (defined in [RFC2046]) for the format.

<DanC> this sentence almost says that follow-your-nose is important. "Web architecture depends on applications having a shared understanding of the messages exchanged between agents (for example, clients, servers, and intermediaries) and a shared expectation of how the payload of a message -- a representation -- will be interpreted by the recipient."

HST: That looks like it just covers the Category A cases (WG owns the media type and is creating it), but not the Category B cases (WG does not own the media type, it already is in use, but not registered)

<DanC> +1 give the AB points 1 and 2 (and perhaps note in passing http://www.w3.org/2001/tag/2004/0430-mime )

NM: Indeed -- Mark Baker asked if we might consider extending that finding to cover [Category B]

VQ: Should we conclude this by just sending points 1 and 2 to the AB?

<DanC> that's reasonable "it's a process question and beyond the TAG's remit to say when latitude is in order"

NM: If that's all we do, we risk the AB understanding this as telling the AB to make it firm and strong
... Maybe we should add something saying that the AB need to understand that sometimes flexibility is required

TVR: Yes, we should make sure they understand that

NM: Can we get an email review of my redraft?

<DanC> I'm willing to be critical path. how about noah sends; if I give a thumbs-up and nobody gives a thumbs-down in 2 days, noah sends to AB

VQ: Need to converge quickly, before next telcon, as AB is waiting
... Let's do this by email

NM: Deadline by Friday?

VQ: Yes

<scribe> ACTION: NM to redraft, forward to AB unless unresolved negative comments from TAG members [recorded in http://www.w3.org/2006/07/25-tagmem-minutes.html#action01]

issue w3cMediaType-1 status

VQ: Does TAG issue number 1 relate to this? There's a two-line resolution, but it still shows as open.

<DanC> Ian, is it really the case that tag issue #1 is still open? is that a bookkeeping bug?

<DanC> re issue #1, http://www.w3.org/2001/tag/2004/0430-mime is approved (or so it claims) and it refers to http://www.w3.org/2001/tag/ilist#w3cMediaType-1 , which doesn't say it's closed.

<Ian> http://www.w3.org/2001/tag/2004/0430-mime

<Ian> "This Finding was derived from discussion of TAG issues w3cMediaType-1, customMediaType-2, and nsMediaType-3 but in some cases extend beyond the specifics of the issue that was raised."

<Ian> And there's this:

<Ian> http://www.w3.org/2002/06/registering-mediatype.html

<Ian> So I believe we have guidelines (now well-integrated into the "how to do a rec-track transition documentation")

<Ian> And the other question of issue 1 was "Should they be defining them at all?"

<Ian> I think the answer was "yes, and that we review that info during rec-track process"

<Ian> I am not sure that there was a formal resolution that the issue was closed, or an attempt to see if Mark Baker is satisfied.

DC: I think the issues list is correct, the issue is still open, the finding didn't close it.

VQ: Well, there's been nothing on this since 2002. . .
... OK, that's alright as it stands then

issue XMLVersioning-41

VQ: Some discussion last week, but document was new at that time
... Any comments after a further week ?

NW: Will send comments in the next week or so

<DanC> (the "Example 2: Evolution of Producers and/or Consumers" diagram is nice.)

<DanC> (indeed, partial understanding is the holy grail.)

<noah> Noah thanks Dave (a lot) for taking the effort on partial understanding. I think it's really the crux of the overall issue. Once we have that right, I think complete understanding follows almost as the degenerate case.

DO: I'm getting value from interaction of this task with similar task for the W3C XML Schema WG
... For example, the extension and restriction constructs relate to the syntax set, not the information set -- that distinction is proving very helpful
... So our work is having an immediate impact outside the TAG

<DanC> (dave, you'll be here on 8 Aug?)

<dorchard> I'll be on 8 Aug

HST: I'll try to review it by the next meeting

<DanC> cool.

DO: I'll try to get another iteration on partial understanding in the next two days

VQ: Yes please

<DanC> (indeed, congrats DO for slaying the CVS dragons and producing http://www.w3.org/2001/tag/doc/versioning )

VQ: Much better now that it's in the same place with the other draft findings
... Could part 2 be moved as well?

DO: Yes, I will do that [Done]

issue URNsAndRegistries-50

VQ: Triggered by email from DC: http://lists.w3.org/Archives/Member/tag/2006Jul/0019.html

<DanC> public mail: http://lists.w3.org/Archives/Public/public-semweb-lifesci/2006Jul/0027.html

DC: Semantic Web Healthcare and Life Sciences IG are discussing LSID: Life Sciences Identifier, used for e.g. proteins -- URN:LSID:domain.name:....
... The usual reasons were given for not using http, and I pushed back
... I've been reading the LSID spec (from OMG) -- there is one new thing, maybe, namely that they are immutable

TBL: What's 'immutable'

DC: The binding between URI and representations is one-to-one
... Also, you have to be able to tell that something is an LSID by looking at it

TBL: You don't get any out-of-band info with a URI

NM: This is different -- this is something they want you to be able to do by inspection
... without doing a GET
... They've deployed software which does retrieval
... Using what protocol?

DC: Not sure -- mag tape?

HST: One of the emails suggested DNS followed by http/ftp/SOAP

TBL: They want flexibility, but not too much, and not yet
... They're still discussing whether an LSID identifies a string of bits or a protein

NM: Is this urgent?

HST: Too late, the spec. defining this is fully baked

DC: Not too late, W3C IG is deciding on Monday 31 July to endorse this as Best Practice or not

<DanC> the monday meeting is a meeting of the group described in http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup

<Norm> Dan asked me about this, I wrote http://norman.walsh.name/2006/07/25/namesAndAddresses in response. FWIW.

HST, DO: DC, could you help us understand the ways in which the draft URNsAndRegistries finding didn't stand up as an argument?

NM: Well, it reads as an argument which works for someone who's already convinced, but won't work for a skeptic. . .

<DanC> http://www.w3.org/2001/tag/doc/URNsAndRegistries-50.html#protocol_independent

TBL: But LSIDs are for retrieval. . .

DC: Still thought there was this thing about using magtape, just associating the name with some of the bits there
... I'm prepared to think of that as caching. . .

NM: Well, that doesn't really fit with the ordinary understanding of caching

DC: I think it does -- the names are distributed, and then wrt those names, the data is distributed (via magtape, because they're big)

TBL: There's this problem with the difference between the URI scheme and the protocol
... 'http' refers to both

<DanC> "I am sure though that you will appreciate that this is not at all the same thing as being able to actively source the named object from multiple places" -- http://lists.w3.org/Archives/Public/public-semweb-lifesci/2006Jul/0032.html I started a "it's not at all obvious to me that these are different" reply, but then hit the immutability stuff in the LSID spec and paused.

TBL: LSID is just a scheme -- maybe it will turn in to a retrieval method in time

DanC, I refer you to the ARK approach (http://www.cdlib.org/inside/diglib/ark/arkcdl.pdf, which uses http and a checksum in the URI

HST: I learned, in writing my replies to the LSID mail, that the burden is on the protocol to specify what schemes it can work with, not the other way around.

VQ: Running out of time -- DC, you have more information for your meeting on Monday?

DC: More friends, anyway

VQ: HST, you have more input to the finding?

HST: Yes

metadata-in-URI

NM: We were nearly ready to go, then in parallel with our decision to finalize we got more feedback from Stuart Williams and Bjoern Hoerrman. ...Formally, I've still been instructed to publish. The TAG therefore needs to decide which if any of new input received merits redrafting. ...Suggestion: I will shortly post two related things. (1) a new draft which reflects changes already requested by TAG but not comments from Stuart or Bjoern -- i.e. the draft I would have finalized if we didn't get the late comments and (2) an analysis of their comments, suggesting which if any I recommend reflecting in yet another draft. ...With that in hand, the TAG can either signal that I should stick with the draft from (1) or do more work.
... The comments are all in the thread from the announcement of the draft on www-tag

<noah> http://lists.w3.org/Archives/Public/www-tag/2006Jun/0157.html

<noah> Above is Bjoern's note on metadatInURI

<noah> Stuart's note: http://lists.w3.org/Archives/Public/www-tag/2006Jul/0026.html

Adjorned