13 Jul 2005


See also: IRC log

1. Roll Call
BEA Systems, Mark Nottingham
BEA Systems, David Orchard
Canon, Herve Ruellan
Iona Technologies, Suresh Kodichath
Microsoft Corporation, Martin Gudgin
Nokia, Mike Mahan
Oracle, Anish Karmarkar
SAP AG, Volker Wiechers
SeeBeyond, Pete Wenzel
Sun Microsystems, Marc Hadley
W3C, Yves Lafon
IBM, Noah Mendelsohn

Canon, Jean-Jacques Moreau
Microsoft Corporation, Jeff Schlimmer
Oracle, Jeff Mischkinsky
Sun Microsystems, Tony Graham
Mike Mahan

Martin Gudgin.


2. Agenda Review


3. Approval of May 18th Minutes
Approved without objection

4. Action items
Yves ACTION: There may be no need for any action. Simon has now sent mail to the comments list
Gudge's ACTION: Done

5. SOAP 1.2 PER Status

W3C Team pushback on Rec20/22
Yves: Team looked at the issue and decided that the approach was not the right one. Want a proposal like the one made around 1 year ago.
... The document around use of XML Schema 1.0 with 1.1 documents could be considered new info
Yves, please can you post links to your e-mail
Gudge: Don't think we should revisit this. We spent a lot of time on it before and the errata has been public for a year.
Mark: Does this require changes to software
Yves: Not aware of anyone enforcing XML 1.0 requirement
Mark: Doesn't media-type restrict to XML 1.0
Yves: <I missed this>
Marc: We would get interop problems if some stacks support XML 1.1 and some don't
Mark goes and looks at the errata
<Yves> http://www.w3.org/2003/06/REC-soap12-20030624-errata.html#E19
Thanks Yves!
Mark: Seems like we're trying to create a profile in the spec.
Gudge; But we tried to do better than a profile last year and didn't get anywhere.
Mark: Are the team saying they won't support the move from PER to 2e (paraphrase)
Yves: If it doesn't pass the director's publication request we'll have to go back to WD
Gudge: I don't understand what WD means for a PER
<mnot> http://www.faqs.org/rfcs/rfc3902.html
Mark: I don't feel strongly about the errata language esp. when the media-type forces us to use XML 1.0
Marc: Why does media-type force XML 1.0
Mark: Quotes from rfc3902 section blah subsection blah
Mike: So does the media-type force XML 1.0 or not?
<Yves> http://www.w3.org/2000/xp/Group/5/02/PER/soap12-part2.html
<Yves> http://www.w3.org/2000/xp/Group/5/02/PER/soap12-part2.html#ietf-draft
Marc: So the proposal is that SOAP messages in the abstract can contain characters that can't be serialized as XML 1.0, but the binding restricts to XML 1.0
Anish: Recollection was that the problem was that the schema was non-normative and we didn't want to move because Schema didn't support XML 1.1
... If the schema was non-normative would that allow us to fix the problem.
<Yves> http://www.w3.org/TR/2005/NOTE-xml11schema10-20050511/
Anish: Are there any constraints that are specified in the schema that are not specified in the spec
Gudge: No
Mark: We don't use any XML 1.1 features in our schema
Anish: This is more than a validation issue. You can serialize using XML 1.1 but you can't use any new features.
Yves: Only XML 1.1 support is for NEL
Gudge: Everyone was happy with that
Yves: I voted against that resolution at the time but didn't raise a formal objection.
Mark: Couldn't we have a normative schema but not require that messages conform to it
... Doesn't require schema validation
Mike; Are you concluding that this note is of no help?
Mark: It seems to be about how to implement an XML 1.1 aware Schema processor
... Doesn't seem to address our problem.
Mike; Noah is invested in this... We should table discussion until he returns next week
Mark: Schema difficulty seems to be the sticking point
Gudge: Sticking point for me is that we're being asked to reverse a decision we took a year ago
Discussion tabled until next week

6. Potential work item: SOAP one-way MEP and Binding
Mike: Various e-mails on this subject. 'paper' trail is in the agenda
Mike; I'd like to hear opinions from the group
Yves: If we decide to do a note, the charter says we can start work on new bindings so a note would be fine.
Yves; If we want a Rec we need to move to the new patent policy. But the old Recs (SOAP 1.2, MTOM etc) will stay under the CPP.
Mike; I asked the CG why, if the requirement was only for a note, why does XMLP need to do it.
Mike; If this group is going to do it, a Rec seems more sensible
Marc: The async task force is still discusing this. And some of the discussion has centered around whether SOAP MEPs are even needed
Yves: Because async-task force has not finished discussing this request is somewhat preemptive
Anish: this work came from the WSD WG, not the async task force
... WSDL has a one-way MEP and SOAP bindings and they're concerned about demonstrating interop
Some discussion of whether we're being asked to do this work or whether we're being asked what we'd do if we were asked to do this work...
Mike: I think we're being asked to do the work
Mark: It's not clear the async task-force is going to come up with any further output. We have a piece of work that will satisfy WSDesc requirements
Mike; So should we be producing a Note or a Rec?
Mike: Don't think anything in the request from WSDesc pushes us in either direction
Anish: Don't think the WSDEsc WG care. Not sure they can normatively reference a note
Mark: Not sure from a process standpoint either.
... Only makes sense to do it here in XMLP if it's a Rec
Marc; We'll need a small but significant change to our binding to support one-way
Anish: LAst time we discussed this there was pushback against doing this work in XMLP.
Mike: Request was for us to do the work and to let the CG/WSDesc know whether we're publishing a note or a Rec.
Some discussion of how WSDesc schedule will line up with our schedule if we decide to produce a Recommendation
Mike: Yves what would it take to produce a new charter? What would we have to produce by which dates?
<dorchard> that was a concern of wsdesc: what if the xmlp rec happens after wsdl 2.0 goes to rec.
Yves: The current charter allows for work to start. Would need to recharter under new patent policy before first publication
<cferris> me s/ere/here
Mike; If we don't have time/resources to meet WSDesc schedule requirements what's the point of doing the work.
<scribe> ACTION: Yves to check process requirements regarding refering to WDs/LC WDs from further advanced documents [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action01 ]
<scribe> ACTION: Yves to talk to Philipe about whether there needs to be an AC Review for a new charter [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action02 ]
<scribe> ACTION: Yves to come up with proposed new charter language [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action03 ]
Mike: Probably want to postpone voting on this until timing WRT WSDesc is clearer
<scribe> ACTION: Mike to check timing WRT WSDesc deliverables. Due for next meeting 2005-7-20. [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action04 ]
Dave: WSDesc group was concerned about timing issues in terms of making sure the WSDL one-way MEP could refer to the SOAP one-way MEP.
... WSDesc plan for CR is fairly soon, say October... We'd need a WD to refer to...
Yves: I have an action to investigate this
Mike: Plan to take a vote on this next week (when Noah is back)

7. Issues raised by WSD/Async task force
<Yves> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Jun/0107.html
Anish summarises his e-mail
<dorchard> Another way of looking at question #2: Can the binding "switch" the MEP from req-resp to soap-resp?
Marc: I think that the binding currently requires a SOAP envelope in the HTTP response
<anish> http://lists.w3.org/Archives/Public/xml-dist-app/2004Jul/0013.html
<anish> http://lists.w3.org/Archives/Public/xml-dist-app/2004Jul/0015.html
<marc> http://www.w3.org/TR/2003/REC-soap12-part2-20030624/#tabresstaterecheads indicates that the response can only contain a SOAP message
DaveO: Another question is what's the wording as it stands vs what implementations use...
... I wonder if impls have treated this as MAY contain rather than MUST contain a SOAP message
Chris: In the context of WS-Addressing if you specify a reply-to are you expecting a response?
... Seems to me that while binding goo cares, applications don't actually need to know
DaveO: Question is do SOAP MEPs live close to the application or close to the binding being used
Chris; There are two types of binding, transfer binding and app level binding
various discussion of whether SOAP is one-way or two-way...
daveo: I believe that if in the years since SOAP 1.2 went to Rec there haven't been any issues WRT 'non-compliant' impls then we should update the spec to match the impls
anish; I agree, i was just concerned with what the spec actually says.
<cferris> +1 to daveo's point
Mike; Take more discussion to mailing list
<cferris> i too am concerned with what the spec says

Summary of Action Items

[NEW] ACTION: Mike to check timing WRT WSDesc deliverables. Due for next meeting 2005-7-20. [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action04]
[NEW] ACTION: Yves to check process requirements regarding refering to WDs/LC WDs from further advanced documents [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action01]
[NEW] ACTION: Yves to come up with proposed new charter language [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action03]
[NEW] ACTION: Yves to talk to Philipe about whether there needs to be an AC Review for a new charter [recorded in http://www.w3.org/2005/07/13-xmlprotocol-minutes.html#action02]