WS Policy Working Group
14 Feb 2007

<PaulC> Agenda: http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0069.html

Secretary for next week's telcon

DaveO: I'll be secretary for next week's (2007-Feb-21) telcon

Ashok: I can be secretary for the 2007-Feb-28 telcon

Roll call

Review and approval of 2007-Feb-07 minutes

Paul: Discussion w.r.t. corrections from IRC log
... Minutes changed to reflect corrections
... Comments on minutes?
... Hearing none, any objection to modified 2007-Feb-07 minutes

February 07, 2007 Minutes adopted unanimously

Future meetings

Paul: I will chair Feb 21, Chris will chair Feb 28

Editorial Team Report

Paul: Prasad presenting report for editors

Prasad: Not much change from agenda

Current Editors drafts:

http://www.w3.org/2002/ws/policy/#drafts

S-Policy Candidate Recommendation Editor's drafts, Editors

http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0208.html

Status: WG members have approved these drafts for transition to CR.

Updated CR drafts, Editors

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0052.html

Updated interop scenarios document, Editors

http://dev.w3.org/cvsweb/~checkout~/2006/ws/policy/interop/WS-Policy-Scenarios.pdf

Prasad: Moving forward from last week's call - topic later in today's telcon

CR interop scenarios and testing

Paul: Pushed this item down in the list - may not get to it today
... Felix wondering if need to discuss target namespace change in target XSD file(s)
... Done right thing by updating target namespaces in XSDs

Review of Action Items

Paul: ACTION-189 Pending
... ACTION-201 still pending?

s/

<whenry> http://www.w3.org/Bugs/Public/show_bug.cgi?id=4253

Paul: 4253 - message 0078; 4255 - message 0076

<whenry> http://www.w3.org/Bugs/Public/show_bug.cgi?id=4255

William: My action is done

Paul: ACTION-201 Done
... ACTION-204 Done
... ACTION-211 Pending
... Under action item added Prasad's email ?ing how we would go about this AI
... ACTION-213 Pending
... WSDL 2.0 fragment ids
... ACTION-214 - my action Done but awaiting response from Ashok

Ashok: ?s about style and format, but I can [proceed] on feature 23

Paul: Ashok if you can track this and formulate ?s against doc in example
... ACTION-216 Pending

Umit: Done but have to send it out to WG

Paul: ACTION-218 Done

<prasad> Regarding AI-211, I am going forward with the direction proposed in my email and I have expanded the scope to include all scenarios under UDDI , expecept 42 that skip is doing. I plan to propose a draft for 42 also. Subject to review by the team.

Paul: ACTION-219 In Progress - received email from Maryann to adjust agenda to discuss
... ACTION-220 Done - revised documents are ready
... ACTION-221 Done

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0037.html

Paul: ACTION-189, ACTION-216, ACTION-219 open

Liaison items

Paul: Feedback to WS-Addressing WG on Web Services Addressing 1.0 - Metadata specification pending
... WS-Policy Call for Implementations pending
... Move on to agenda item 7 (per revised agenda)

WS-Policy Call for Implementations

Paul: the Working Group SHOULD include a report of present and expected implementations as part of the request
... IBM, MSFT, WSO2 and Layer7 confirmed participation
... Anyone else confirm that they can participate

Ashok: Yes, ORACLE will participate

Symon: Yes, BEA as well

Prasad: Yes, WEBM as well

Umit: Are the exit criteria we're discussing as proposed by Chris?

Paul: Reiterated what we discussed last week in 10) iii) in agenda
... Include report of present [and ...] implementations
... Seven companies participating
... Two companies interested in participating to demonstrate features relating to UDDI (thanks Prasad and Skip Snow)
... Questions is whether these companies have software that provide WS-Policy (since that would be our CR draft)
... Next item: the Working Group MAY identify specific features of the technical report as being "features at risk"

Asir's note:

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0022.html

Paul: No one suggested to add items beyond what was in Asir's note
... Those three items therein qualify as features at risk?
... Make a decision which of these belong in "at risk" category
... Criteria: If we don't put a feature at risk and cannot demonstrate interoperability, to remove the feature would require going back to WG
... Discussion on list - attaching policies using UDDI
... What is WG's position

Toufic: Point out that there are a couple impls of the attachments/UDDI that i am aware of - WEBM and [...]
... My organisation is more than willing to demonstrate interop for that feature

<GlenD> Commentary - we as a group should commit to what we specify, and I believe none of these things should be "at risk", else we should have already removed them from the spec. Leave them in, and if indeed we don't get interop, doing one more LC cycle isn't so horrid.

Paul: 2 orgn's are Systinet->Mercury->HP and WEBM
... And as identified by WG, Amberpoint

+1 to GlenD

Toufic: Metadata attach at present based on previous spec rev; we also may use some of our own namespaces, but in essence this is WS-Policy support for such attachments
... Any comments on that?

<umity> +1 to GlenD

Paul: Would be best to have demo of UDDI attaching W3C Policies

Chris: Some discussion off list w.r.t. WSDL 2 - given that there are only 2 orgs participating in WSDL 2 CR - are we holding up our spec because of the higher bar for the WSDL 2?
... Plans from some of the other participants - need clear understanding if we can get 4 impls of the WSDL 2 binding

Paul: Don't want to say WSDL 2 is at risk - but want to be practical as to number of impls and our schedule for getting through CR

Prasad: Question w.r.t. Agenda 10iii
... Mark all at risk or can the WG apply a lower bar?

Paul: We *can* apply a lower bar but need to explain that decision
... Lack of impls would be a valid reason

Prasad: UDDI applies as special case

<toufic> +1

Paul: Set lower bar for WSDL 2 and UDDI?

<cferris> +1

<umity> +1

+1

Ashok: Last version of Oracle product had external policy attachment - used old WS-Policy spec - different syntax - whether or not we can actually interop on that will need to be thought about, but the feature is there

Paul: If we try and use shipping product that refers to prev version, we're walking a thin line - not what the W3C would like to see

<toufic> agreed

<fsasaki> agreed

Paul: Participating in CR would mean using W3C WS-Policy namespace

Umit: No one is talking about ignorable attribute - if we don't have enough impl experience to get everyone lined up in time - why can't we apply the same reasoning as done with WSDL 2 and UDDI

<GlenD> We discussed ignorable last time and on the mailing list, FYI.

<GlenD> There was some pushback against marking it at risk.

Umit: Propose that WG consider setting exit criteria for all features except for interoperable impls

Skip: +1 to Umit's proposal

<maryann> +1

<maryann> +1 for IBM

Proposal: Four specific features - agree not to have at risk - set lower bar for interop in order to get out of CR
... all other features would need to be supported in the 4 interop impls

<whenry> +1 for IONA

Paul: Pragmatic approach toward CR

<toufic> +1

+1 from Adobe

<Ashok> +1

Paul: Any objection to putting no features at risk and for item 10iii identifying interop criteria

<cferris> proposal: PR transition criteria is as follows: All features of the spec will demonstrate 4 or more interoperable implementations with the following exceptions: WSDL2.0 attachment, External Policy Attachment, UDDI attachment and wsp:Ignorable, each of which shall require 2 or more implementations.

DaveO: Why not place one (or more) features at risk? Don't have to go back to WD if only one impl shows up
... UDDI and WSDL 2 seem to be features we could put at risk
... Not a lot of downside to put feature "at risk"; upside is we can always cut them if can't impl them in time

<cferris> I would point out that there was CONSIDERABLE pushback on placing UDDI atachment at risk

DanRoth: Some interop for external attachments - do people know we'll have interop for this? I've heard this for UDDI, WSDL and one person speak to external attachments.

Ashok: Ship product code, but supports pre-W3C Policy; won't be able to do interop next month, but it is possible in May 2007, will inquire within Oracle

Paul: Anyone else able to deliver interoperable software for WSDL 2 or other at risk features?

Maryann: We at IBM can

Paul: DanRoth nodding head
... Anyone cannot live with Umit's proposal?

<cferris> proposal: PR transition criteria is as follows: All features of the spec will demonstrate 4 or more interoperable implementations with the following exceptions: WSDL2.0 attachment, External Policy Attachment, UDDI attachment and wsp:Ignorable, each of which shall require 2 or more implementations.

Paul: Any one cannot live with this proposal?

<PaulC> a) Ignorable Policy Assertions b) External Policy Attachment c) Attaching Policies Using UDDI d) Attaching Policies Using WSDL 2.0

<prasad> Who did we have committed for wsp:Ignorable?

<PaulC> Proposal is CR exit criteria is 4 implementations for all features except the four above for which it is 2 implementations.

Paul: Any objections to this?

<cferris> RESOLUTION: WG unanimously agrees to the proposal at http://www.w3.org/2007/02/14-ws-policy-irc#T17-48-39

RESOLUTION: PR transition criteria is as follows: All features of the spec will demonstrate 4 or more interoperable implementations with the following exceptions: WSDL2.0 attachment, External Policy Attachment, UDDI attachment and wsp:Ignorable, each of which shall require 2 or more implementations. CR exit criteria is 4 implementations for all features except the four above for which it is 2 implementations.

Paul: Covers 10ii and 10iii
... Item 10iv

The announcement MUST indicate a minimal duration, before which the Working Group MUST NOT request a Call for Review of a Proposed Recommendation; this minimal duration is designed to allow time for comment.

Status: members were asked to consider the end of May (following the May F2F) as the earliest possible date to transition to PR.

Paul: Any objections that we don't go to PR before end of June 2007?

<cferris> +1

Paul: March F2F, May F2F, then June to clean up data we have to present for PR transition
... No objections

RESOLUTION: End of June 2007 to be earliest date of transition to PR

Paul: 10v - The announcement SHOULD also include the Working Group's estimate of the time expected to gather sufficient implementation data.

<cferris> RESOLUTION: WG unanimously agrees to go with end of june as earliest date at which it will seek transition to PR

Paul: Draft call for implementation to be put together, all docs ready to go, deliver to team today

<cferris> congratulations are in order for the WG!

WSDL 1.1 Element Identifiers

"One small step for [a person], one giant step for [humankind]"

Paul: Discussed syntax for some WSDL 1.1 identifiers:

Ashok's proposal:

http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0169.html

DaveO's correction:

http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0180.html

At the F2F meeting WG members asked for more time to consider this matter. There are two alternatives to choose between. We have had to delay this item several times due to lack of key participants.

DaveO can live with either version of change syntax

<cferris> Ashok's confirmation of daveO's correction: http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0181.html

Ashok: Still want to see change made

Maryann: No objection to change

<Yakov> no objection

Paul: Anyone else who supports change proposed by Oracle
... Any one objects to this change?

Silent majority wins

<cferris> RESOLUTION: 4251 closed with proposal from daveO http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0180.html

<scribe> ACTION: Editors to implement above resolution (http://www.w3.org/2007/02/14-ws-policy-irc#T17-57-45) [recorded in http://www.w3.org/2007/02/14-ws-policy-minutes.html#action01]

<trackbot> Created ACTION-222 - Implement above resolution (http://www.w3.org/2007/02/14-ws-policy-irc#T17-57-45) [on Editors - due 2007-02-21].

<fsasaki> ACTION: pcotton2 and cferris to get explicit comment from WSDL 2 WG on status [recorded in http://www.w3.org/2007/02/14-ws-policy-minutes.html#action05]

<trackbot> Created ACTION-223 - And cferris to get explicit comment from WSDL 2 WG on status [on Paul Cotton - due 2007-02-21].

Guidelines issues, Part 1

ISSUE 4035: [Guidelines] Section 2 should account for interop impact of non-wire or one-party assertions and ignorable property

http://www.w3.org/Bugs/Public/show_bug.cgi?id=4035

http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0007.html

ISSUE 4040: [Guidelines] Update guidelines to include discussion of ignorable, Frederick http://www.w3.org/Bugs/Public/show_bug.cgi?id=4040

http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0025.html

ISSUE 4262 Use of @wsp:optional and @wsp:Ignorable on t...

http://www.w3.org/Bugs/Public/show_bug.cgi?id=4262

See Monica's recent email:

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

Paul: Issue 4035: Section 2 of Guidelines - needs to account for interop impact of non-wire or one-party assertions and ignorable property

<PaulC> http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0007.html

<cferris> ACTION: Felix to update WG public page with WSDL1.1 Element Identifiers ED link [recorded in http://www.w3.org/2007/02/14-ws-policy-minutes.html#action06]

<trackbot> Created ACTION-224 - Update WG public page with WSDL1.1 Element Identifiers ED link [on Felix Sasaki - due 2007-02-21].

Paul: Adopt this (http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0007.html) to take to the WS-A WG
... Give people opp to review proposed text and see if they agree with the proposed change

<dorchard> why not wait for the update to the doc based upon the issue and then send that to WSDL WG?

http://dev.w3.org/cvsweb/~checkout~/2006/ws/policy/ws-policy-guidelines.html?content-type=text/html;%20charset=utf-8

<cferris> will the link change?

<TRutt__> I like it

<dorchard> yes, the TR version won't change

Paul: No one on queue - affirmation?

Monica: "Behaviour" and "Interoperability" used interchangeably in the proposal - was that intended?

Umit: No

<dorchard> just update the HP?

Monica: Revision on text?

<dorchard> home page?

Umit: Pertains to WS-A Metadata spec - use ignorable for assertions that will impact the wire - anonymous required
... Using optional and ignorable for anonymous required

<whenry> +1 to Umit

+1 to Umit

DaveO: Haven't reviewed text but like the gist of it

Paul: Nothing about this actually in our framework document?

Umit: Not quite

Paul: Defn of ignorable policy assertions - may be ignored for policy intersection

Maryann: Impact on interoperability

<whenry> +q

Paul: If apply [Dan's] interpretation to Framework, is that what Frederick's test intends to achieve

DanRoth: Yes - there is some clean up needed
... Text is good, can we discuss now to clean up or go to the list?

<whenry> -q

Paul: Can we at least minute observation about significance/use pattern for ignorable

Maryann: Need to respond to WS-A WG in timely manner

Paul: Easy to respond in LC comments to say "using ignorable incorrectly" in timely manner

<whenry> +q

<danroth> Ignorable assertions are assertions that may be ignored during intersection.

<danroth> Intersection is used to test compatibility for interop.

<danroth> Assertions that are relevant to interop should not be marked as ignorable assertions.

<whenry> People on quue?

<PaulC> For cases where the provider does not intend the assertion to impact interoperability it may mark it as "ignorable".

<PaulC> For cases where the provider intends the assertion to impact messages on the wire and interoperability, it MUST not use "ignorable".

<GlenD> s/it may/they may/ ?

<SergeyB> +q

+1 to Umit

TomRutt: Looking at 1/31/2006 version of WS-A Metadata?
... Need to look at that one.

<danroth> Can someone send a link to the version Tom is referring to?

TomRutt: Only wording that bothers me is "avoid..."

<PaulC> Metadata spec: http://www.w3.org/TR/2007/WD-ws-addr-metadata-20070202/

<danroth> The problem is in example 3-12

Umit: "ignorable" used with assertion - semantically affects interop - user of nested assertion with "ignorable" is not what was intended; our observation is still valid

TomRutt: Agreed with words we're penning now

<cferris> I would like to point out that the WSDL metadata nested assertion in question (wsam:AnonymousResponses) DOES NOT carry any semantic that impacts interoperability

Maryann: Put something before working group - summarise WG perspective - hence close bugs

<whenry> Still on the queue

<Fabian> yes, I did lament!

Paul: Move on to other two items in agenda item 9?

Notes that we still have a queue

William: Give WS-A WG guidelines of intent?
... Umit's point: When I read this I see lot of very well defined standards legalese - can we provide a simple version of intent?
... Ask that w.r.t. wire behaviour, is there any case we could put something on the wire that could be ignored anyway
... clear understanding of intent of what s/b on the wire

<Fabian> stirred and fried :-)

Sergey: General advice?
... Thinks that's strong for general advice

chris: Not certain I agree with analysis expressed with WSDL WG Metadata spec having any impact on interop
... WSDL metadata nested assertion in question (wsam:AnonymousResponses)
... In context of "ignorable" - anon responses and non-anon responses do not impact anything on the wire

Umit is next on the queue

<Zakim> umity, you wanted to respond to Tomm and to

<cferris> fwiw, I agree with the guidance that has been proposed

Umit: Reason for interop - if have assertion in place, change or put something on the wire to indicate that (and how) using semantic in that assertion
... Implicitly affects wire interoperability - if logic is flawed, wire interop needs to be further qualified

<cferris> my response to Umit is that the semantic implied by wsam:AnonymousResponses is NOT a requirement

<cferris> it does NOT impact interoperability

Paul: Don't want to debate concepts here

<cferris> it is only a hint to the client that maybe, it might want to think twice before sending a message that has a non-anonymous response

Umit: There also is the question of the document design itself - wanted to look at our semantics and make them clear for our response to WS-A

<cferris> however, my key criticism is that there is NO WAY for the wsam:AddressingRequired assertion to carry a semantic of ANY requirement as to what the client MUST or MUST NOT do on the wire

Fabian: Second statement put down is too strong (For cases where the provider intends the assertion to impact messages on the wire and interoperability, it must not use "ignorable")

<cferris> +1 to Fabian's suggestion of SHOULD vs MUST

Prasad: Thinks "MUST" is more appropriate

<umity> Chris, I think we are on the same page.

+1 to Fabian's SHOULD

<Zakim> dorchard, you wanted to say we can just point them to the message and say we'll do something like that

DaveO: Approach - point WS-A to message (vs. document) if not done in time?

<cferris> my point though is that the criticism of their use of wsp:Ignorable on the nested asseertion is actually misplaced BECAUSE the nested assertions do NOT place any requirement on the messages exchanged

Paul: "...Some WG members think X..."

<umity> That is part of the other feedback that I am currently crafting.

<cferris> the key text in the WSAM spec is as follows: "The absence of the wsam:AnonymousResponses policy assertion within a policy alternative does not indicate that the endpoint will not accept request messages with response endpoint EPRs that contain the anonymous URI as an address; it simply indicates the lack of any affirmation of support for anonymous URIs."

TomRutt: Assertions you may want to mark "ignorable" - some special cases pointed out by WS-A WG that don't want to mark them as "ignorable"

<cferris> there is no way that they can express NOT

<cferris> absence means nothing

<umity> we are well aware of this issue

TomRutt: Close to last example but may need mods

<cferris> my personal opinion is that the entire nested policy section of their spec is severely flawed

Paul: Look at related issue 4040

Umit: Issue 4040 - related to 4035, if close 4035, close 4040

<PaulC> Issue 4040: http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0025.html

<PaulC> Issue 4040: http://lists.w3.org/Archives/Public/public-ws-policy/2006Dec/0025.html

Monica: Other discussion on list w.r.t. usage of "ignorable" which impacts this

<PaulC> Issue 4262: http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

Paul: Move on to 4262

"The framework specification does not explicitly state if an assertion can be

marked both optional and ignorable. However, as we discussed since

@wsp:optional is just a syntactic simplification, it is permitted to mark an

assertion with both the @wsp:optional and @wsp:Ignorable with the value of

"true" for both."

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0017.html

<monica> This also applies to my comment in #4040 (but can be handled in #4262). http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=4262

<PaulC> Can/should you use wsp:ignorable and wsp:optional on the same policy assertion?

<PaulC> The current Framework doc does not appear to prohibit use of both wsp:ignorable and wsp:optional on the same policy assertion.

Paul: Any disagreement on that statement?
... Prasad's statement - maybe we should make this clear in Guidelines
... For some assertions it may be useful?

Ashok: In some cases this is useful
... This is ok

DanRoth: Agree that the Framework allows it and Ashok's interpretation
... should - it would be a strange policy to have both ignorable and optional on the same policy assertion

<umity> +1 to Dan

<whenry> +1 to Dan. Do we have an example?

DanRoth: I take it to mean, "I may [...] your message"
... I take it to mean, "I may log your message"

Paul: Can someone describe where WS-A is using both on the same policy and where/why it may be wrong?

TomRutt: 2 use cases: 1) client of WS-A wants to ensure that policy assertion "anonymous" is there - they would not use lax algorithm - for it to work, endpoint would need optional; 2) client doesn't use or care about anonymous, lax algorithm would/could be used, and ignorable is needed for that use case

Prasad: Difficultly we have with ignorable is that we don't have concrete examples, doesn't illuminate use case for using together with optional

Umit: Tom has clearly identified why the use cases in the WS-A MD spec - but don't use both ignorable and optional together - my conclusion w/b that the example would need anonymous with optional and ignorable

<Fabian> http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

Fabian: Look at http://lists.w3.org/Archives/Public/public-ws-policy/2007Feb/0061.html

Monica: Applicable to Issue 4040

Fabian: Cases exist where it makes sense to use both optional and ignorable

<monica1> When a policy assertion can not be marked as optional (or is not marked as such) and Ignorable is used, that assertion is not>> optional (is required) for a client that does understand it.

<monica1> When strict mode is applied for matching, Ignorable exists on compatible assertions.

Fabian: First rule doesn't apply to this case

<monica1> The intersection algorithm allows the client to filter out assertions that it does not understand and that were marked Ignorable. This is the mustUnderstand inverse.

<monica1> After intersection the resulting policy could contain assertions marked with Ignorable and the resulting policy is applied to the messages. Those assertions that the client understands are not ignored.

Fabian: With optional assertion, lose information; with both, unless use lax alg or otherwise filter it out, keep information

<TRutt__> I agree with UMIT. In fact the example with ignorable was put in at the 11th hour. A better example which would satify both use cases would be to put both optional and ignorable on the anonymous assertion. We should provide such feedback to the ws -addressing group.

chris: Want to be clear that there's nothing in WS-A MD spec about using them together, but they can be used together as there's nothing in Framework to explicitly prohibit it; as such it can be done and s/b ok

<GlenD> yup

+1 to Umit and TomRutt

Paul: Shouldn't necessarily derive feedback to WS-A WG that presupposes that their use is incorrect

TomRutt: WS-A's weird use cases do warrant both optional and ignorable together, or teaching them how to properly use nesting - w/b satisfied by the first (put both optional and ignorable in)

<cferris> I should be clear that I don't think that they have a problem with use of nested assertions.

<cferris> my criticism is that their assertion definitions SHOULD convey a requirement

Prasad: Clarify that we want to provide guidance when to use both optional and ignorable together and/or not

<umity> +1

<cferris> they SHOULD NOT convey "support" as that semantic implies no requirement

Paul: Think we need an email dialogue on the list w.r.t. this

Umit: Will meet coming monday to discuss among task [sub]group for this

Paul: Last comments on this?

chris: Round of applause to WG for reaching this significant milestone

<fsasaki> applause to!

chris: Special thanks to editors for going above and beyond on their tasks

<fsasaki> thanks to Prasad esp. as well

Telcon adjourned

<whenry> \quit