Tag teleconference of 26-April-2005

26 Apr 2005


See also: IRC log


Roy Fielding, Vincent Quint, Tim Berners-Lee, Noah Mendelsohn, Norm Walsh, Ed Rice, Henry Thompson (most of meeting)
David Orchard, Dan Connolly (joined near end of meeting)
Vincent Quint
Noah Mendelsohn


Tag teleconference of 26-April-2005

<scribe> scribe: Noah Mendelsohn

<scribe> scribenick: noah

Future Meetings

May 3rd: no regrets for now

May 10th: Norm and Tim will be unavailable

<timbl> I note May 10th is the WWW conference in Chiba Japan

May 17th: Tim at risk

For next week, scribes will be: David Orchard if available, otherwise Norm Walsh

Approval of previous minutes

Review of: http://www.w3.org/2001/tag/2005/04/19-minutes.html

Norm: haven't reviewed but no problem if others agree.

RESOLUTION: Minutes of April 19th accepted (http://www.w3.org/2001/tag/2005/04/19-minutes.html

Henry reports that he has been on phone for past 5 minutes

TAG Publications and "Heartbeats" (Includes extended RDDL discussion)

Norm sent a note (member only) on errata: http://lists.w3.org/Archives/Member/tag/2005Apr/0036.html

VQ: do we have errata?

NW: yes, 3 or 4 reported to comments list, all editorial

<scribe> ACTION: Norm to gather errata list for consideration next week. [recorded in http://www.w3.org/2005/04/26-tagmem-irc]

NW: RDDL to do may be mine. It had history with Tim Bray, then Paul Cotton, both now gone.
... I've somewhat lost track of where it stands, but I'll take the lead if we figure out what we want to do.
... The AC asked the TAG to produce a rec-track document, presumably based on RDDL.

VQ: Do we have a draft?

NW: No.
... I'm willing to produce if asked to do so, either based on RDDL 1.0 or something else. What do we want to do?

<ht> Would prefer 1.0

NW: ht has pointed out use of RDDL 1.0 in some W3C specs. Microsoft is also deploying.

TBL: Are you saying TAG should produce a Rec?

NW: AC asked us to.

TBL: GRDDL affects this, no?

HT: Solves a different problem.

TBL: Why?

HT: RDDL solves: what should you do specifically with namespaces. GRDDL is about: "I've got a doc that serves two purposes (1) to be an XML document and (2) how to harvest metadata"

TBL: GRDDL allows you to extract (RDF?) information from something that is not in RDF form.

NW: One of the objections to earlier Namespace formats was that they were not in RDF and should have been. GRDDL takes the heat off of that.

HT: Some of the problems we're solving don't need all the RDF mechanism.

TBL: Yes, but for the many namespaces built for RDF it makes great sense.
... For a semantic web language, then OWL should be the language of the namespace document.

HT: Historically, we decided early that there was no single thing you wanted to say about a namespace.
... Among the many such things is what the RDF schema is, what the XML Schema is for the serialization, etc.

TBL: I understand. I think you need to divide the cases. XML Schema makes no sense for RDF vocabularies.

HT: But with GRDDL it does make sense. It gives me license to author in colloquial XML a document that is known to contain RDF statements.
... For such a document, all of that is pertinent.

TBL: That makes sense, except that when I extract the RDF information I would not expect the extracted doc to be in the same namespace.

... Still, my point is that the original RDDL was based on the presumption was that there would be more than one document of interest, and that it would contain pointers to one or more things as necessary.

TBL: Well, only if we say that this is only for non-semantic web namespaces.

HT: I'm less convinced than you that the differences matter, but I can easily live with a statement that at least for now, the RDDL approach is aimed at non-semantic web namespaces.
... Information please: does OWL say use it as a namespaces doc?

TBL: No, but the TAG did some sort of best practices note (can't find it right now) which says to use OWL for namespace docs, and I think it's common practice.

HT: OK, so finding might say: if (isSementicWeb) then {use owl} else {use RDDL}

NW: I'm not convinced that using RDDL for all is bad, with the RDDL pointing to the OWL. I want human readable documentation, etc.
... I'm thinking of Ed Dumbill's work on software project descriptions. It's both RDF and not.

<ht> DOAP

NW: I don't totally disagree with Tim, but I'm not totally convinced either.

TBL: what's he really done?

<Roy> pointer to RDDL in minutes?

<ht> http://usefulinc.com/doap

NW: A constrained RDF that's usable as XML.

<ht> http://www.rddl.org/

TBL: right, that's the tricky case
... specifically because there's an issue of whether the fragid identifies the concept or the markup

HT: argh, not httpRange-14!

NRM: No, smaller than httpRange-14. We need to ask: "what's the media type?" in this case, and what answer is given to Tim's question about fragids?

<Roy> http://www.w3.org/TR/2004/REC-webarch-20041215/#namespace-document

RF: We've been around this before. Key question is: what's the requirement for the document? I think we had consensus on one thing, which is that the key requirement is that it be human readable.

<Norm> Roy is right, we did come to consensus on that point

TBL: I'm not sure we said that. For some cases, machine readable is important.

<timbl> I don't think we had a consensus that human readable was primary. Tim Bray said so strongly. I am prepared to defend strongly the semantic web use of OWL and RDF in general.

RF: I mostly don't want to go around same ground again. Ideally, it's useful for human purposes and machine readable as necessary.

TBL: We have a current suite of specs that tell a semantic web agent how to go out and find semantic web information. This would require a new spec and engineering into future software. Is this a good thing to require for agents?
... In fact, you can use a stylesheet to make an owl doc human readable

HT: We enabled indirections such as RDDL for locating W3C XML schemas from the namespace URI.

<Norm> If the cost of the redirect is too expensive, then the cost of the retrieval is probably too expensive too.

HT: The rec was written to allow you to implement indirection through things like namespace documents. Most implementations do the right thing when a RDDL doc comes back.
... This is a relevant precedent. It was easy and lightweight. About 10 lines of code in the validators I built.

TBL: Most semantic web software has no XML processing included, except for limited purposes.

RF: They're not extensible for new media types?

TBL: There's no XML schema processing in the RDF processors.

HT: You've misunderstood. I'm only drawing a parallel. I'm pointing out that the fact that it was easy to add to XML schema processors suggests it will be equally easy to add such indirection to RDF processors.

TBL: Adding something simple to a system that's clean and coherent results in a kind of complication.
... e.g., using different DTD syntax for XML complicated everything
... Even using the XML stuff for RDF was a complication, but it would be good to hold the line there. OWL is in RDF, so we can get up into a pure semantic web world quite quickly and cleanly.

<Zakim> noah, you wanted to ask how sem. web namespaces are the same and different from others

<Zakim> Roy, you wanted to ask if there are any examples of such an OWL namespace description to look at

<ht> I always thought it was the XSL WG's requirement that produced Namespaces. . .

NRM: Tim tells us that sem. web. is a clean closed consistent world, but some of the complication comes from the fact that it shares a namespace mechanism with a wider XML world.
... This allows a nice "pun", in which the serialization of an RDF namespace is the corresponding (same) XML namespace. We also can reference XML datatypes.
... So, complication will pop up with either choice we make: if we want the sem web side to be clean and consistent, then the story about namespaces as a whole gets more complicated, because we have a two part rule. If we have a uniform story about all namespaces (e.g. RDDL), then the sem. web side gets a bit more complicated. Take your choice.

<timbl> I have :)

<timbl> (taken my choice)

<ht> DanC, are you happy with the XRI response?

<Norm> http://localhost:8123/2005/04/26/atom.rdf

<Norm> http://norman.walsh.name/2005/04/26/atom.rdf

<DanC> I think so, ht. the one I reviewed and commented on was pretty close, and I assume the subsequent edits remain in that neighborhood

NW: Well, if we need a special case for something like semantic web, I'm not sure it's worth doing at all.

VQ: Time's about up. I had hoped we could decide about publishing something based on RDDL? Looks like 'no'.
... Let me ask the original question again: is there anything we could publish as a heartbeat from the WG?

TBL: Note or working draft?

VQ: Working draft would make requirement heartbeat. Do you have any in mind, Tim?

TBL: No. Norm said he would write something.

Dan appears to have joined the call.

NW: I'm reluctant to write draft without consensus on what high level approach to attempt.

TBL: We're being lured into grabbing an answer that's lying around. This sounds like a job that's more suited to a WG that would focus on it more seriously. When we started this, RDDL looked pretty well baked.

DC: We have to publish something or close the group, or get permission from the director to continue anyway. We don't have to decide what it is today.

VQ: My understanding is RDDL is not ready for pub soon, and I'm asking if we have anything else to publish.

TBL: What about asking Norm to write a finding discussing the RDDL/OWL questions in relation to namespaces?

<DanC> (pls let the record show we're discussing issue namespaceDocument-8, not just the heartbeat publishing requirement)

<scribeEditorialInterjection> See topic title for this section of the minutes.

NW: I could pull together a short finding talking about earlier RDDL work, and today's discussion.

TBL: Would OWL be in there?

NW: Yes, if it belongs at all. I know where I'd like it to come out, but this finding is where it all belongs.
... Will try to do something this week.

DC: You're going to hit all kinds of fun stuff, such as the fact that rddl.org is a non-w3c basis for a namespace name.

VQ: what's on the agenda was Dan's proposal to publish something based on RDDL.

DC: That was an aside.

HT: I could possibly move along the finding on issue 50, which would give us something to publish.

<DanC> (what HT is drafting on issue 50... doing that as a /TR/ publication would be different from the way we've done findings in the past, but not a bad idea)

VQ: That would give us a first WD

HT: Right, that's what I meant.
... Do findings count as heartbeats?

DC: No

HT: We need discussion on that meta topic.
... We are not working primarily in a Rec mode, and should not do backflips to fit into a process not tuned to our needs.

VQ: Yes, we should discuss that.

<timbl> TBL: I agree with HT

<DanC> (I don't mind changing the rules. My job as team contact is to enforce them or change them. I'm happy to do either.)

NRM: At next meeting?

VQ: yes.
... Henry, can you make progress on issue 50 anyway?

HT: Yes, I intend to try.

Planning next F2F meeting in Boston

VQ: We started some discussion about what we should do as next technical document. We agreed to do some preparation in advance, but agreed that the main decision making should be at the face to face.
... This is a reminder to start that discussion.
... Ed, was there something else you wanted to ask?

ED: Partly travel plans, partly wanting to focus on issues.

VQ; Meeting will start morning of June 14. Some of us will leave early- to mid-afternoon of 3rd day.

VQ: Please send travel plans to tag mailing list.

<Roy> 9am

VQ: Start time 8AM, Tues, 14-June.

HT: I'd prefer we go as late as possible on the 3rd day.

No conflicts reported.

VQ: Since there are no conflicts, we'll go a full day on the 3rd day. So, the meeting will be three full days, the 14th 15th and 16th of June, going until late afternoon on the 16th.

<DanC> (not sure that was really 8am and not 9am, but probably doesn't matter)

(right, we can fine tune later.)

<Zakim> Roy, you wanted to suggest we have a F2F goal for a REC track webarch volume 2 that contains only an outline

NRM: I think priority and as much time as necessary should be given to figuring out what our big themes are for the coming year or two. Issues should be discussed, but only time permitting.

RF: We could try to publish an outline for Web Arch 2. That would take most of 3 days.

<Zakim> noah, you wanted to suggest main focus should be on trying to identify what our big themes will be for next year+

<DanC> (yes, let's work on an outline; dunno if I want to publish it)

<Zakim> DanC, you wanted to ask about xquery namespace study, and timing

<ndw> DanC, please send those as LC comments!

DC: I spent a lot of time working bottom up on the functions and operators stuff.
... I think I'm suggesting the group spend some time working through those particular details.

VQ: Let's wrap up the agenda discussion. I think it will be useful to have a draft agenda early. I will aim for one month in advance, which is mid May.

<scribe> ACTION: Vincent to prepare by mid-May a draft agenda for the June face to face meeting [recorded in http://www.w3.org/2005/04/26-tagmem-irc]

<ndw> We've hit my hot list for the f2f: outline WebArch V2, httpRange-14, and the issues DanC mentioned

Feedback on XRI proposal

ED: We've had some discussion on tag@w3.org. What's the right order for putting something on www-tag and/or sending to Oasis?

Discussing Henry's (member only) draft at: http://lists.w3.org/Archives/Member/tag/2005Apr/0062.html

DC: we need better links to XRI docs.
... Actually, this is close enough.

TBL: This is good, but somewhat underrepresents level of detail of our analysis. Much better than nothing.

HT: Should we point to the httpDNS thread in the public mailing list?

<DanC> for names consisting of an adiminstrative hierarchy and a path, HTTP/DNS is as good as it gets

HT: I feel we should reproduce the tag@w3.org discussion on www-tag@w3.org
... I can send a package with all the emails tomorrow.

<DanC> PROPOSED: to respond to XRI docs ala 0062 in the tag archive, plus public version of technical details discussed in tag@w3.org ...

HT: I can then reference that and other threads in the formal response.

<DanC> PROPOSED: to respond to XRI docs ala 0062 in the tag archive, plus public version of technical details discussed in tag@w3.org by HT, unless show-stoppers are raised within 2 business days

<Ed> +1

VQ: OK, and formal response can go out by Friday after we review
... who should send?

<DanC> mechanics: VQ mail it to www-tag and then copy it into the OASIS form.

TBL: Anybody on TAG can do as long as they correctly speak for the TAG.

HT: I think it would be more polite from VQ

DC: The mechanics of getting feedback to them are more complex than you'd like.

TBL: Do email first so at least we have web archive copy for reference.

<DanC> mechanichs: VQ send to xri-comments@lists.oasis.org, cc www-tag; then take a pointer and put it in the OASIS form

NRM: Suggest that Henry's note warn that we owe feedback at end of week, and that we need to focus discussion toward that goal.

General agreement to that suggestion.

DC: Silence is assent.

<DanC> RESOLVED: to respond to XRI docs ala 0062 in the tag archive, plus public version of technical details discussed in tag@w3.org by HT, unless show-stoppers are raised within 2 business days

VQ: Right, if there's no objection, I'll send Friday France time.

Note that the HTTPDNS thread (note) mentioned above is at: http://lists.w3.org/Archives/Public/www-tag/2005Apr/0076.html

Binary XML

VQ: I see some progress in email.
... Ed, do you feel your action is fulfilled?

ED: The working group doesn't exist, where do comments go?

DC: Mailing list exists after group is gone. Use mailto:public-xml-binary@w3.org.

VQ: Do we expect to do more?

<timbl> http://lists.w3.org/Archives/Public/www-tag/2005Apr/0059.html

DC: Where's the list?

<ndw> I agreed with Dave's points, it's a good list

<DanC> http://www.w3.org/2001/tag/2005/04/26-agenda.html

Discussing Ed's additions to the list at: http://lists.w3.org/Archives/Public/www-tag/2005Apr/0085.html

VQ: Proposal for today is to ask everyone to check this message, and see whether we think this is our reply.

<Zakim> DanC, you wanted to re-iterate Orchard's concerns about explaining how http/DNS addresses these issues and to

NRM: Should we also get ready to signal whether and how we would contribute to the possible chartering of a new WG?

TBL: That's the director's decision.

NRM: I think it's for the TAG to decide whether they do or don't want to provide input to the director and/or the AC. I'm suggesting it may be constructive to do a round of thinking now about where we stand on that.

DC: Do we have a draft response.


ED: I can try to draft one.

DC: I had said we should try to make a decision on issue 30 before June AC meeting. Because the team may not make that deadline anyway, the need for feedback is less urgent.

VQ: Ed's help in refining this is most welcome.

ED: Input solicited.

NRM: Please extract from email thread as well.

ED: Will do.


VQ: Two or three items remain.
... Roy, could you look at putMediaType-38?

RF: I've looked, just have to write the email. Put on agenda for next week.

<Roy> I've looked, just have to write the email.

HT: I sent some email on httpRange-14.

VQ: Please send input for next agenda before next Monday, France time.

Summary of Action Items

[NEW] ACTION: Norm to gather errata list for consideration next week. [recorded in http://www.w3.org/2005/04/26-tagmem-irc]
[NEW] ACTION: Vincent to prepare by mid-May a draft agenda for the June face to face meeting [recorded in http://www.w3.org/2005/04/26-tagmem-irc]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.112 (CVS log)
$Date: 2005/04/29 01:32:33 $