TAG Teleconference, February 14, 2005

14 Feb 2005



+1.916.785.aaaa, Norm, DanC, [IBMCambridge], Noah, Hoylen, Vincent, Roy_Fielding, +1.856.914.aabb, Ashok, TimBL, dezell, Mary_Holstege, Michael, DOrchard




<Vincent> Yes, I have invited all the bots

Thank you!

<Norm> noah, Vincent: I'm waiting for the heating company to arrive, our furnace isn't working, so I'll be interrupted when they do arrive. (It's 60 degrees and falling in the house)

<Vincent> hi ed

<Ed> hello

Norm, OK: we're sharing scribing anyway. I'll be sure to cover when you leave. Is it supposed to be over 60 in the house?

Roll call, agenda review

<scribe> scribe: Noah


From the tag, we have: Noah, Vincent, Roy, Tim BL, Norm, Ed Rice

Regrets from: Henry Thompson, David Orchard

<Norm> I'd like to see xml:id/C14N discussions on the agenda for next week

From schema WG: Hoylen Sue, David Ezell, Mary Holstege, Michael Sperberg-McQueen


<Ashok> Noah, I'm also here

We will have a telcon next week on the 21st

Ashok Malhotra is also here from the Schema WG

Dan Connolly will scribe next week.

Review of Minutes for 2/7

<DanC> (ed sent something thursday?)

<DanC> http://www.w3.org/2005/02/07-tagmem-minutes.html

Should we accept minutes of 2/7 that were released on Thursday: http://www.w3.org/2005/02/07-tagmem-minutes.html

Dan: there is some miscellaneous stuff in there that should be cleaned up.

ATTENDANCE: David Orchard joins the call

We will not accept 2/7 minutes now. Will wait for cleaned minutes.

<scribe> ACTION: Ed Rice to clean up 2/7 minutes and either email or commit somewhere in CVS

<DanC> (2/7 is not I18N-happy. it's 7 Feb)

ACTION 1=Ed Rice to clean up 7 Feb minutes and either email or commit somewhere in CVS

<DanC> (or 2005-02-07)

Telcon Scheduling

Vincent proposes 1PM Eastern Time for future TAG Telcons

Various members assent.

<DanC> 22 Feb

Vincent: We have decided to schedule TAG telcons on Tuesdays at 1 PM Eastern time, starting Tues 22 Feb

Tech Plenary

<DanC> ACTION: Vincent to arrange bridge for 1pET tuesdays starting 22Feb

Noah sends REGRETS for 22 Feb.

<timbl> Tim bl sends regrets for Feb 22

Dan: I was asked for input by Paul Downey

Vincent: work directly with the organizer if you have questions or concerns about the panel
... Noah will be on future of XML panel

Norm: I think I'm on the XML futures panel, if it's the same as the XML 2.0 panel

<DanC> (it's not called "XML 2.0" anywhere that I can see, norm. "Session 3: Where XML is Going, and Where it Should (or Shouldn't) Go" Moderator: Rich Salz. http://www.w3.org/2005/03/02-TechPlenAgenda.html )

Noah: FWIW, I don't think I'm specifically representing the TAG on future of XML, was invited because of lightening talk I gave last year on XML 1.1 issues.

Vincent: we have been solicted to propose lightening talks at the Plenary. Any candidates?

OK, nothing specific on lightening talks here.

TAG F2F at Plenary

Vincent: We have TAG meeting on Monday morning of TP, with WS-Addressing meeting at the end of that.

VQ: will work on agendas etc. tomorrow morning. Suggestions welcome.

TAG Liaisons during Plenary Week

See: http://www.w3.org/2001/tag/2005/01/TechnicalPlenaryLiaisons.html for list of proposed meetings

<Roy> I am available for all liaison meetings listed.

VQ: Propose Thurs. 11 AM for Core WG meeting.

Noah: Doublebooked with Schemas

Roy is available for all liaisons.

Norm will be there, and is on core.

Ed can attend.

VQ: OK, we'll confirm that we will meet with them.

<scribe> ACTION: Vincent to confirm Thurs meeting at 11AM with Core WG

VQ: Proposal for joint meeting with XML Schema WG Thurs after lunch

Dave Orchard will be there

Norm: would like to be there, but may conflict with core

Noah: will be there, on both groups

<Roy> ditto

Vincent will be there

Roy will be there

VQ: Resolved, we will meet with XML Schema on Thurs at 2 PM

<dezell> sorry missed the opportunity -- for how long? 1 hr? 2?

VQ: Proposal to meet with QA WG. Discussion with chairs suggests best course is joint calls after the plenary.

<DanC> (I'm bummed about the scheduling of the TAG/WS-Addressing and TAG/CDF meetings. sigh.)

(I'm bummed Dan will miss WSA)

VQ: Compound Document group proposes to meet with us on Monday at 3PM

<Norm> I've stepped away to deal with the heating guy. brb.

Dan says he's chairing other WG's Mon & Tues, how about Thurs.

VQ: Well, we already have 3 meetings on Thurs., could be hard

Dan: OK, do it without me. Too bad.

<scribe> ACTION: Vincent to figure out whether and when we want to meet with Compound Documents group

VQ: If you are interested in CDF meeting and have scheduling constraints, send me mail by tomorrow, 15 Feb.

Extensibility and Versioning Discussion with XML Schema WG

VQ: Dave, as editor of draft finding, please give us a TAG status report

DO: the draft finding has been considerably refined from 2003 version.

See http://lists.w3.org/Archives/Public/www-tag/2004Nov/att-0071/versioning-part1.html and http://lists.w3.org/Archives/Public/www-tag/2004Nov/att-0071/versioning-part2.html

DO: Discusses some of the decisions a language designer would want to consider, and gives advice on making those decisions.
... Sample issues: do you want compatible or incompatible extensions, should owners of other Namespaces be able to extend your language.
... which versions of schemas do producers and consumers have?
... how does a consumer know which particular version a particular component is part of?
... are you using new namespaces for new constructs? Spectrum of possibilities.
... Compared to 2003, trying to provide somewhat more formal definitions of terms like language.
... That was part 1. Part 2 discusses schema-specific issues, and in this case mainly W3C XML Schemas.
... This cut focusses on W3C XML Schemas. Plan is to include RDF, OWL and Relax NG. Norm has been helping me with Relax NG. Hope is to put that in soon.
... Plan to spend a bit more time on this now that I'm back on the TAG

VQ: Would someone from schema give us an update on your work.

David Ezell will be known as DE.

DE: We've been looking at this for several years.

<DanC> (this use cases thing is great... I only discovered it minutes before the telcon, though. has everybody else read it in detail? http://www.w3.org/XML/2005/xsd-versioning-use-cases.html)

DE: Our mandate is to improve things in XML Schema 1.1
... XML Schema 1.0 was focussed more on expressing constraints. In Schema 1.1, we plan an explicit focus on frameworks for extensibility and versioning.
... Schema WG has posted a number of documents including
... Use cases http://www.w3.org/XML/2005/xsd-versioning-use-cases.html

<Zakim> noah, you wanted to update status a bit

<Zakim> DanC, you wanted to ask if the problems with extensibility also relate to having a PSVI

Noah: yes, just reminding that I am still working on my action to help Dave O. with higher level issues in the draft finding

Dan: does having a PSVI make it harder? I think some early Schema WG members thought so.

<Zakim> noah, you wanted to comment on psvi

<DanC> -> An Approach for Evolving XML Vocabularies Using XML Schema http://lists.w3.org/Archives/Public/www-tag/2004Aug/att-0010/NRMVersioningProposal.html

<Zakim> MSM, you wanted to say that in fact I think the PSVI is the key to support for versioning

<DanC> (what I hear is yes, folks are looking at interactions between PSVI and versioning)

DE: I can't say much officially on behalf of the wg. PSVI is a fertile area for discussion.

Noah: actually, there have been proposals to leverage the PSVI to help applications discover which content was truly expected (validated against explicit declaration) vs. being tolerated for forward compatibility (validated against some sort of wildcard or extensibility hook)

<DanC> ("the finding"=??)

Michael Sperberg-McQueen: MSM

<Zakim> dorchard, you wanted to talk about parsers impl of partial validation

MSM: I don't share Dan's concern. I think indeed that David's draft findings fail to fully explore some of the positive potential of the PSVI.

DaveO: One reason draft finding doesn't talk much about partial validation is that in practice today's processors don't report the PSVI in that detail.
... I was trying to talk about solutions that were practical today.

MSM: one of our requirements (specifically RQ-144) for version 1.1 for XML Schema is to clarify the conformance requirements for how much of the PSVI processors must support.
... We are seriously considering but not yet committed to also naming useful combinations so a processor could claim: "I report the Red subset."

<dorchard> BTW, when I specification "optional", I mean both optional in the spec and optional in the impl. Amazingly, in general, not all software implements all required parts of specs.

MSM: you can find pointers to proposals from the Schemas editorial page.

<Zakim> dezell, you wanted to report on PSVI support

FYI: link to definition of schema requirement 144 is http://www.w3.org/XML/Group/2002/07/xmlschema-1.1-current-reqs-list.html#WhichPSVIPropertiesReqd (member only)

DE: At our schema meeting at Redwood shores many implementors were present, and all claimed to support full PSVI.

DaveO: For the information of new tag members, I've been doing a fair amount of the work on the finding.

VQ: How can we best coordinate with the Schema WG on this?

DaveO: for a start, I think the scenarios (what I've called mine) and use cases (what the Schema WG has called theirs) are important
... we need to understand, beyond use cases, which parts of this work the schema WG thinks is its responsibility

<MSM> dorchard: getting agreeement on the scenarios / use cases is really really important. Extensibility is not JUST a question of the schema langauge, or the schema, but systemic

DaveO: we need to work together to figure out what behavior should be from a web arch perspective.
... having the schema use cases public helps.
... then there the nuts and bolts of what happens in Schema NG (presumably Schema 1.1 [scribe]) is mostly the responsibility of the Schema WG
... I think the mandate I've received is to explore many different schema langs.

<Zakim> dezell, you wanted to embellish

DE: we agree very much that the use cases are important.
... we've tried hard not to be self selecting in use cases, I.e., to not focus only on things we think we know how to solve or should solve on our own.
... we shouldn't prejudge use cases as to which are the business of W3C XML Schema, per se.
... it's in our charter to make sure that XML Schema 1.1 has a much better story on versioning.

<DanC> (our list of actions is a mess)

<DanC> (but it's worth noting that an action is continued explicitly)

DE: I keep hearing form people in (my employer's group), the National Assoc. of Convenience stores, that they expect XML to make things simple, but versioning is a big challenge for them. Users do care.

VQ: Noah, what was that you volunteered to do?

Noah: to indeed fulfill the action I was assigned to work with Dave to explore some higher level issues for the draft findings.

VQ: new action?

Noah: No, I think it's in the list somewhere. If not, we should add it, but I won't generate a duplicate now.

VQ: Anyone with anything else on this topic?

MSM: Yes. I think you're focussing on older processors/applications getting data conforming to newer schemas.

Seems that you're worrying mainly about the case where the old processor can't get at the new schema and resulting PSVI.

We think that there are use cases where you can get the new schema and use cases where you can't.

DaveO: it seems to me the 80/20 case is where the software that did not evolve is not configured to pull new schemas or similar descriptions dynamically. Could happen, but not 80/20.

MSM: are you saying what they have done or should do?

DaveO: security models are a factor.

MSM: we may need to agree to disagree
... I think there will be situations where people should fetch the new schema

DaveO: agreed in principle, but I'm not sure it's practical in enough cases to be worth major focus

MSM: Another point. You seem to be assuming that applications will necessarily roll over and die when data is not valid per a schema. That's only one model. Many applications will fall back after a validation failure.
... so, there's a model in which the early schemas do not try to partially validate future extensions, but in which the applications have some other model as to how to proceed when the validation doesn't fully succeed.

DaveO: I don't understand.

MSM: I'll try again. The draft finding conflatest the set of documents that a given application will accept with those that are valid per its (preferred) schema.

<MSM> the finding seems to me to conflate two distinct notions: (1) the set of documents valid against schema v.N and (2) the set of documents application software will accept and process.

DaveO: I think you're saying there can be implicit schemas which are larger than (in your example) or smaller than the explicit schema written in a specific schema language.
... I agree with that. Have been trying to figure out how to articulate that.

<MSM> And it's important to note that it's not really hard for parser-generator technology (and thus for code generators) to generate a parser that is more forgiving, more capacious, and accepts partially valid or invalid documents

Tim: would the broader schema be for what the application accepts in its "longer lifetime", I.e. for all the future variants that the application will to some degree tolerate?

Dave: Java code could add constraints

<MSM> tim, in the ideal case, yes, that larger 'set of all future documents' is what one might plausibly want the application software to accept.

Tim: I'm raising the case where constraints are looser than you might think.

<Zakim> DanC, you wanted to suggest findings or chapters list in http://lists.w3.org/Archives/Public/www-tag/2005Feb/0093.html

Noah(nonscribe) notes: we also have to watch for the fact that sometimes people introduce intentionally incompatibilities. Sometimes the v8 flavor of a language is intentionally incompatible with v1 of an application, because we know very old versions are not in use any more.

<DanC> "Extending and Versioning XML Languages Part 2: Schema Languages"

Dan: I think your title in part 1 sounds broader than what you actually discuss.

DaveO: originally, this was going to be a one parter on using XML schema language to extend and version. People came back and said, please split it into a general and a specific.

Dan: Norm wrote "consider the case of XSLT 2.0". That's a good idea, and I have a list of about 10. Here goes:

<DanC> [[

<DanC> -- XSLT extension functions

<DanC> -- adding <img> to HTML

<DanC> -- adding <form> to HTML

<DanC> -- adding <frames> and <script> to HTML

<DanC> -- SMIL system: extensions

<DanC> -- adding rdf:parseType="Collection"

<DanC> -- not adding rdf:parseType="Quote"

<DanC> -- adding Content-Transfer-Encoding to HTTP

<DanC> -- the emergence of TLS in HTTP (new URI scheme, new port; good

<DanC> idea or not?)

<DanC> -- CSS forward-compatible parsing rules

<DanC> -- SOAP must-understand

<DanC> -- the XPointer scheme registry

<DanC> ]]

Dan wonders whether 3 or 4 of the topics listed above would be a good focus?

<DanC> -- http://lists.w3.org/Archives/Public/www-tag/2005Feb/0093.html

DaveO: yes.
... for example, XSLT is used as an example of multiple versions using the same namespace, with a version attribute used to distinguish processing semantics.

<MSM> MSM thinks "adding 'script' to HTML" should also be considered, because it's a case where the HTML extensibility rules were not quite what people would have liked to have

DaveO: some people want it bigger, some smaller

<DanC> (did I forget <script>? I meant to include it, MSM)

<Zakim> DanC, you wanted to say that issue41 has always felt like a book, to me; that extensibility _is_ W3C's core competency, and that if the TAG spent a year exploring it, that would be

<Zakim> dorchard, you wanted to respond to Noah "issue 41 navel gazing"

Noah: I think the TAG should focus on the higher level issues that bear on the health of the Web, the schema WG should focus on XML Schema mechanisms, and somewhere in the middle are XML details that aren't validated by schemas.

Dan: If the TAG's next major work was 50-100 pages on versioning, that might not be a bad thing.

DaveO: the age old question of going top down or detail-oriented.
... I think architects must focus on the details to be successful.
... I campaigned in part on having the tag focus on the details of this particular issue.

VQ: We're running out of time. Thanks to all!

I have made the request to generate http://www.w3.org/2005/02/14-tagmem-minutes noah

Summary of Action Items

[NEW] ACTION: Ed Rice to clean up 2/7 minutes and either email or commit somewhere in CVS
[NEW] ACTION: Vincent to arrange bridge for 1pET tuesdays starting 22Feb
[NEW] ACTION: Vincent to confirm Thurs meeting at 11AM with Core WG
[NEW] ACTION: Vincent to figure out whether and when we want to meet with Compound Documents group
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.111 (CVS log)
$Date: 2005/02/14 21:32:38 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.111  of Date: 2005/02/12 19:57:54  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/Topic: Topic/Topic: Tech/
Succeeded: s/Vincent/VQ/
Succeeded: s/Sample/DO: Sample/
Succeeded: s/tolerate./tolerate?/
Found Scribe: Noah
Inferring ScribeNick: noah
Default Present: +1.916.785.aaaa, Norm, DanC, [IBMCambridge], Noah, Hoylen, Vincent, Roy_Fielding, +1.856.914.aabb, Ashok, TimBL, dezell, Mary_Holstege, Michael, DOrchard
Present: +1.916.785.aaaa Norm DanC [IBMCambridge] Noah Hoylen Vincent Roy_Fielding +1.856.914.aabb Ashok TimBL dezell Mary_Holstege Michael DOrchard
Agenda: http://www.w3.org/2001/tag/2005/02/14-agenda.html

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: ed rice vincent

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)

[End of scribe.perl diagnostic output]