W3C | TAG | Initial agenda | Previous: 17 Jun | Next: 1 July

Minutes of 24 June 2002 TAG teleconference

Nearby: Teleconference details issues list www-tag archive

1. Administrative

  1. Chair: NW. Scribe: IJ
  2. Roll call. Present: PC, NW, TB, DO, RF, CL, IJ. Regrets DC, TBL, SW
  3. Next meeting: 1 July. Regrets: DO, PC, NW
  4. Resolved: Accepted 17 June minutes confirming acceptance of both augmentedInfoset-22 and xlinkScope-23.
  5. Confirmed status of completed actions

1.2 Completed actions?

  1. IJ 2002/06/17: Add augmentedInfoset-22 to issues list. Assigned to Tim Bray
  2. IJ 2002/06/17: Add xlinkScope-23 to issues list. Action IJ: Ask TBL to take ownership of this issue.
  3. NW 2002/06/17: Call for initial review on www-tag of "TAG Finding: Consistency of Formatting Property Names, Values, and Semantics". Done
  4. PC 2002/06/17: Convey the desire to the XMLP WG that MIME type registration be included in soap spec before going to last call.
  5. IJ/PC 2002/06/17: Update finding to ensure that it's clear that the registration must be part of the document at last call if the WG expects to skip Candidate Recommendation. (Revised finding). Confirmed changes to "Internet Media Type registration, consistency of use" ($Date: 2002/06/24 22:17:30 $).
    CL: I can live with that, though I predict that people will find adding normative parts of the spec after last call is not acceptable.
  6. ACTION DO/TB/CL 2002/05/05: Pending XMLP response, polish up DO's .1-level draft and find out what's going on with XForms

    Done. This action was partly completed and partly subsumed.

  7. charmodReview-17: Confirmed that this issue is closed. Action IJ: Close this issue in the issues list, referring to comments from NW to I18N WG.

2. Technical

  1. Architecture document
  2. Qnames as identifiers
  3. Status of discussions with WSA WG about SOAP/GET
  4. Postponed

2.1 New issues?

None.

2.2 Findings in progress, architecture document (45min)

See also: findings.

  1. Comments on XMLP actions regarding SOAP and MIME type registration. See email to TAG. Report back to XMLP WG.

    Action IJ: Ask Stuart to send a thank-you to the XMLP WG from the TAG.

2.3.1Architecture document

  1. ACTION IJ 2002/03/18: Integrate/combine one-page summaries (Revised 7 June)
  2. ACTION TBL 2002/05/05: Negotiate more of IJ time for arch doc
    NW, TB: We prefer not to close this action item yet until further discussion with TBL.

[Ian]

TB: Next step is to publish this draft of the Arch document. It's well-enough cooked that we should point people to draft document in progress.
IJ: I can live with more wide distribution if we loudly announce that it's a moving target. I'd prefer to wait a tad bit longer.
NW (speaking as NW): I'm inclined to agree with TB.
TB: How about that we agree that this will be published by 1 July (with or without improvements)?
IJ: I can live with that.
TB: Status section should be constructed carefully:
  1. This document doesn't represent TAG consensus
  2. It does however represent a lot of TAG input.
  3. It's a moving target.
  4. [Also, remove the "@@" explanation.]
TB: I would echo comments made last time. When you want to put explanatory text in, put examples instead. In 1.2, "An HTTP URI identifies a document"
RF: That was TBL's view. I'm vehemently opposed to the idea that the URI identifies a document.
TB: I support RF on that.
Resolved: Instead of presenting RF's text as editorial Note, give two views equal footing. Tie to issue httpRange-14 (not exactly that issue, but related).
TB editorial notes:
  1. Hyperlink in scheme property 1 is busted
  2. Principle in section 1.4.1 doesn't have a seq #, it should be 5

IJ: What term should "take precedence" in this document? URI or URI Reference?

TB: Depends on what you're doing. But "All important refs identified by URI", not URI reference.
IJ: I will fix this.
CL: As soon as you have content negotiation, you need to know mime type of response before you interpret fragment id's.
IJ: Could someone write a paragraph on URIs and URI references?
RF: Do you want the politically accurate view or the technical view?
TB: I think both are required to understand what the right thing to do is.
(IJ: I will also look at DO's comments on the arch doc.)
Action RF: Write a para on URIs and URI references.

RF: I will try to have this for next week, but likely not ready by then.

Digression into discussion of revisions to URI spec

[Ian]

RF: I am working on revising the URI spec right now, with Larry Masinter and TBL when he has time.
TB: Why are you editing it?
RF: Integrating corrections, inclusion of IPV6 format, inclusion of some I18N work (but unclear how much). Discussion will take place on uri@w3.org. This week's a good time to bring forward your burning issues on URIs.

IJ: Suggest alerting chairs that this work going on.
RF: I will suggest that to Larry. TBL could also do this.

Returning to architecture document

[Ian]

TB: Important to point out that the term "URI reference" conflates relative URIs and fragment IDs.

RF: Additional BNF terms is one suggested improvement to the URI spec.
TB: That's how namespace names got to be URI references. DC said we couldn't make up a new construct that wasn't in the RFC...

2.3.2 Qnames as identifiers

  1. NW 2002/06/17: Call for one-week review on www-tag of QNames as Identifiers. TAG expects to confirm completion next week. Done

[Ian]

NW: I would like to delay this decision this week, in order to reply to comments from Rick Jelliffe. Rick pointed out that some vocabularies use different mechanisms for associating URIs with prefixes. The finding doesn't consider that usage.
CL: The schematron use of namespace bindings is a type of escaping mechanism.
RF: I agree with CL.
TB: The way that schematron does this is elegant and good.
NW: I think what CL makes sense, but that's not my understanding of the example RJ posted. I thought he wanted to refer to a namespace expression, but instead of declaring with an xml:ns attribute, he used his own element. Therefore, I assumed that when he loaded that data model, I thought he wanted the binding to be understood even though this is done in a proprietary manner.
CL: It's reasonable to expect a schematron processor to understand it, but not a general XML parser.
TB: Rick says "Schematron seems to violate recommendations 1, 4 and perhaps 5...."
NW: The source of my concern is that I thought that one intended outcome of this finding was to make namespace usage more apparent to an XML processor in general. Maybe that wasn't the purpose of this finding.

Action NW: Follow up on Rick's comments/proposal by next week.

2.3.3 Status of discussions with WSA WG about SOAP/GET

  1. ACTION DC 2002/06/10: Send note to WSA WG expressing concern about normative binding for GET.
[Ian]
NW: Where are we on this issue?

DO: I spoke with some of our developers about WSDL. Didn't send DC's note to the WSA WG. Some discussion last week at end of call.: I posted some text today to tag@w3.org about problems I see. I think it's not as clear cut about what the right thing to do is.

TB: I'm beginning to think that TBL was right that we should be worried about this issue (SOAP binding). WSDL has a way to declare a SOAP message available through GET. But doesn't use the "?" syntax to do so. I haven't seen an attempt to harmonize SOAP changes in 1.2 and WSDL for how to do this.
DO: There's no example in the WSDL spec to show how this could be done. This could be done - ask WSDL folks to look at GET example in SOAP 1.2 primer.: Also question of schema for URI-encoded parameters.
TB: Maybe we don't have to figure out what to do. The WSDL guys are the experts. Does it suffice to point out to them that:
  1. Some substantial changes to SOAP 1.2
  2. There doesn't seem to be a way in WSDL to declare this (and no examples illustrating this).
  3. Ask them to make necessary changes.
DO: How do you define type information associated with a URI query string? Doesn't seem related to Web Services (a Web thing, not a Web Services thing). Another way is to define a mapping between schema and URIs to allow you to auto-generate a document that you can validate against schema. I think the TAG should have an idea about a direction to take. In WSDL, the SOAP binding allows you to define types (through Schema).: In WSDL, you define messages (that use types). You define a port type that accepts certain message types. Then you define an actual port (and bind concept into URI). There's also an HTTP binding in WSDL (bindings between types and URIs). But you cannot, e.g., say "?stocksymbol=foo". You can only say "name value pairs after "?" and that's it.: Can't say anything about query string. Missing link between contents of a query string (what names, and values can be, which are required, etc.) and a schema type. No way to associate types and content of a query stringl
[Chris]
In other words, query strings are unstructured.
[Ian]
RF: I agree that the [scribe missed] binding sucks. It does not do what any reasonable programming environment would want it to do. Unclear about what data is acceptable, and format of data exchanged. There are many different ways to do this.
DO: Should XML Schema folks look at this problem? Or HTML WG (since name/value pairs part of HTML spec)?
RF: Doing this is the only reason WSDL exists. WSDL defines an interface and translates programmatically into an application. If it wants to deal with the Web side of Web services, should be a way to address interface. I agree that the one defined by WSDL so far isn't expressive.
CL: If you have a name value pair list, you can go to a flat XML doc easily. If you go the other way, it's not as trivial.
[PaulC]
Note that Noah Mendelsohn has sent a note on this topic.
[Ian]
CL: I suspect Schema WG will balk at being asked to address this. You could have processors take a query string and convert to XML doc (or vice-versa) and use XML Schemas.
NW: I don't see any reason why XML Schema is only way to solve problems like this one. Having said that, DO said something about "slash-separated" stuff....
DO: In WSDL, GET binding where URI components are separated with slashes. Pre-query string, however. SOAP 1.2 primer example used information after "?". You can't do types on parts after "?".
PC: Have you seen Noah's most recent response?
DO: Yes.
PC: He is saying that this is no longer a SOAP 1.2 issue.
PC to DO: Why does the TAG have to do either the WSA's work or the Web Services Description Language WG's work?: Sounds like WSDL WG will encounter same problem and will turn to SOAP 1.2.
DO: My reading of the tea leaves is that the TAG considers this an important issue. I think TAG considers this a high priority for WSDL as well. Also educational backgrounder in architecture of Web Services.
PC: I would paraphrase action item "Because TAG thinks that this is important, we need to ensure that the Web Services groups are aware to changes in SOAP and that additional changes may be necessary."
[TBray]
Do we have anyone on the TAG who's also on WSDL group?
[Ian]
DO: After looking at WSDL spec, seems like schema on query strings a more general problem.
NW: I have appreciated DO's explanation and feel this has changed direction slightly since issue arose. I think we need to communicate with Web services folks to say "Central issue is here."
TB: I think we should liaison with Web Services Description Language WG.
NW, TB: Let's ping WSDL WG to ensure that this is on their radar.
DO: We can ask them to make this a high priority in their schedule.
TB: I suggest we ask DO to talk informally with someone from WSDL WG and come back and reassure us and say they're on it. Or ask us to send a formal message.

Action DO: Contact WSDL WG about this issue to ensure that it's on their radar.

Postponed

  1. Internet Media Type registration, consistency of use
    1. ACTION DC: research the bug in the svg diagram.
    2. ACTION NW 2002/06/24: Create a PNG version of the diagram.
      CL: I suggested fixing this so it had no errors, then creating a PNG version.
  2. augmentedInfoset-22
    1. ACTION DC 2002/06/17: Talk to XML Schema WG about PSVI. Report to tag, who expects to decide whether to add as an issue next week. Done.
      PC: Since I won't be here next week, I would prefer that this issue be on agenda after 1 July.
  3. RFC3023Charset-21
  4. Status of URIEquivalence-15. Relation to Character Model of the Web (chapter 4)? See text from TimBL on URI canonicalization and email from Martin in particular.
  5. If we get here: httpRange-14, namespaceDocument-8

Ian Jacobs, for TimBL
Last modified: $Date: 2002/06/24 22:17:30 $