WS Policy WG
9 May 2007

 

 

<Yakov> scribe: Yakov

agenda http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0098.html

<monica> note: Regrets from Fabian; wave from Java One.

Review and approval of WG minutes

<cferris> http://www.w3.org/2007/05/02-ws-policy-minutes.html

minutes approved

<cferris> RESOLUTION: minutes from 05/02 approved as posted

Future WG meetings

next week - Paul will chair

Abbie needs email addresses from the group members

Paul reminds about the F2F registration

Paul asked to send regrets

<cferris> please visit the registration page and indicate whether you will be attending or sending regrets

<TRutt__> Jacques Durand will not attend, he is the Fujitsu alternate

<cferris> please have him send regrets

Chris - F2F in Dublin - please register

Editorial team report

<prasad> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0079.html

<cferris> the link above is the latest draft of the primer

<whenry> Priceless

Prasad: making a good progress on the Guidelines

Paul: the doc will be on the F2F agenda

Review action items

Chris: ACTION-279 is in progress

Paul to resolve liaison issue

Maryann asked to move the date for AI-286

and add to the F2F agenda

<cferris> ACTION: Paul and Chris to add agendum to cover the discussion around action 286 [recorded in http://www.w3.org/2007/05/09-ws-policy-minutes.html#action01]

<trackbot> Sorry, amibiguous username (more than one match) - Paul

<trackbot> Try using a different identifier, such as family name or username (eg. pknight, pcotton2)

<fsasaki> ACTION: pcotton2 and Chris to add agendum to cover the discussion around action 286 [recorded in http://www.w3.org/2007/05/09-ws-policy-minutes.html#action02]

<trackbot> Created ACTION-296 - And Chris to add agendum to cover the discussion around action 286 [on Paul Cotton - due 2007-05-16].

AI-289: Abbie to update F2F

ACTION-290 Ashok to open an issue

<scribe> done by Ashok

Chris: ACTION-291 is done

Chris/Monica: ACTION-292 is pending

<monica> thanks dave

Chris/Asir: ACTION-293 is done

Chris asked Ashok to open an issue regarding ACTION-294

Asir: ACTION-295 is pending, will be ready for the next telcon

What is the Vocabulary of an Intersected Policy, Ashok

Bob: The concerns from WS-Addressing group have been removed

Ashok: asks about policy about empty nested policy and whether client and server can communicate

David: the answer is no but it may be a flaw in our spec

David brigns up the concern about the scalability, wild cards

Tom: clarifies the empty/intersection issue

the algorithm should be fine from the WS-Addressing perspective

Thanks Chris!

David: if the policy has restrictions, the restrictions have to be listed - and the scalability issue may be a result

Monica: need more discussion

<monica> Monica: Need to consider if we are explicit about this in the Framework. We should be explicit. If not here, in the Primer. Use Addressing as an example.

Dan: responds to Ashok about two policies' compatibility and intersection
... explains section 3.6.1 in WS-Addressing about compatibility

Having compact form of policy expression should address the scalabilty concern

David: responds on the compact form.

Dave's concern is verboseness of the nested case

Monica: brings up the domain specific processing

Paul: Is this AI 286?

Maryann: to extend the AI 286 to accomodate the Monica's proposal.

Monica: do we need to update the Framework about the non-matching case
... there is a confusion about the intersection.

Chris: should the Ashok's example be added to the list of expamples?

<monica> thanks asir

<cferris> ACTION: Asir to provide an proposal that adds an example to 4.5 Policy Intersection that calls out the non-matching case from Ashok's email [recorded in http://www.w3.org/2007/05/09-ws-policy-minutes.html#action03]

<trackbot> Created ACTION-297 - Provide an proposal that adds an example to 4.5 Policy Intersection that calls out the non-matching case from Ashok\'s email [on Asir Vedamuthu - due 2007-05-16].

Paul: is there anybody with the proposal?

Chris: asks people to bring up concerns whether the proposal is sufficient

Paul: asks David to open an issue about potential scalability problem

<cferris> ACTION: David to open an issue related to the scalability of intersection in the face of nested policy [recorded in http://www.w3.org/2007/05/09-ws-policy-minutes.html#action04]

<trackbot> Created ACTION-298 - Open an issue related to the scalability of intersection in the face of nested policy [on David Orchard - due 2007-05-16].

I actually see Bob in the users list

<cferris> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0055.html

<toufic> i apologise but i have to get off the call

<toufic> but i have to moderate a panel

<toufic> and tony is a panelist, so i'll ask him about the F2F

<toufic> bye

<whenry> Regrets I must drop off too.

<scribe> scribe: Yakov

<GlenD> What Chris is saying seems fine WRT behaviors that introduce a /requirement/ on an interaction. For behaviors that simply add extra functionality or introduce options, I don't see a problem with them being active even though they aren't explicitly called out in an assertion.

<fsasaki> scribe: fsasaki

<paulc> Chris's proposal: [Definition: A policy alternative is a potentially empty collection of policy assertions.] An alternative with zero assertions indicates no behaviors. An alternative with one or more assertions indicates behaviors implied by those, and only those assertions. No other behaviors are to be applied for the alternative.

paulc: is this the text of cferris proposal?

<paulc> Chris's proposal also includes the markup in the attachment to his email.

<asir> +1 to chris

chris: yes. It only covers the framework

<paulc> Chris's proposal also requires changes to the other specs.

<GlenD> I don't like the idea that someone can complain that I'm not being a good policy citizen if I throw an optional (non-MU) SOAP header in a message I send.

<Zakim> dorchard, you wanted to say what I think positions are, and what pros/cons of each are.

<daveroth> it looks like WS-Policy attachment doesn't use vocabulary terms at all

dorchard: chris is favoring one position. I want to understand what the positions are

<paulc> Chris's proposal is in http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0055.html

dorchard: I see three positions, I posted a mail a while ago

<monica> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0085.html

(mail from david with the positions)

<daveroth> Please note my response to Dave's summary in message 85

<daveroth> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0086.html

[AIN Vocabulary flavour: Any assertion not in a vocabulary should not be

applied (Original chris proposal)

AIN Closed favour: Any assertion not in an alternative should not be

applied (revised chris proposal)

AIN Removal: Any assertion not in alternative means nothing. It may or

may not be applied.]

<paulc> 1. AIN Vocabulary flavour: Any assertion not in a vocabulary should not be

<paulc> 2. AIN Closed favour: Any assertion not in an alternative should not be

<paulc> 2. AIN Closed favour: Any assertion not in an alternative should not be applied (revised chris proposal)

<paulc> 3. AIN Removal: Any assertion not in alternative means nothing. It may or may not be applied.

<paulc> David is asking if the three above items are the three choices.

<paulc> 1. AIN Vocabulary flavour: Any assertion not in a vocabulary should not be applied (Original chris proposal) 2. AIN Closed favour: Any assertion not in an alternative should not be applied (revised chris proposal) 3. AIN Removal: Any assertion not in alternative means nothing. It may or may not be applied.

paul: who in the WG has proposed no 3? question to dave

dave: I think Ashok, Fabian, Dale
... I seem to see some support of removal of AIN removal

<GlenD> I would support the WG seriously considering removing AIN.

<asir> Dave's characterization of position 2 AIN closed flavor is INCORRECT

<monica> asir please get on the queue

dave: the AIN closed flavor has a downside in composition cases like rsp / rm

ashok: I said that I want the third option and that I liked Monicas wording

see http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0083.html

paul: is your mail message 83?

monica: yes

ashok: talked to our product guys, they said they want the third option

<paulc> Glen - are you going to be in Ottawa?

frederick: brief comment, no mail associated:

<GlenD> Paul - It does not look like it, unfortunately :(

<asir> Ashok mentioned a proposal from Monica .. can't see it in the archive tho

<paulc> Glen: can you then submit your regrets via the registration system?

<GlenD> Absolutely, Paul.

<monica> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0057.html

<fjh> two concerns:

frederick: you could do rm using a policy, and security using no policy. We cannot say everything about behaviors

<fjh> first: statement "no other behaviours to be applied" may be too broad. What if you have policy that only specfies RM. Someday later add Message security without updating policy.

<fjh> Why not possible. Shouldn't additional behavours out of policy be possible?

<GlenD> Yes, IMO

<paulc> Using David O's 3 choices. The following is Monica's text for #3:

<paulc> "When a policy assertion is absent from a policy vocabulary (See section 3.2, Web Services Policy 1.5 - Framework), a policy-aware client should not conclude anything (other than ‘no claims’) about the absence of that policy assertion."

<fjh> 2nd concern: without access to provider policy, but only intersected policy, how can negation be known or understood. Better to say what mean?

<GlenD> ESPECIALLY ones that don't cause any required behaviour

<dmoberg> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0105.html

<monica> monica

dale: I talked in my mail interoperability concerns mentioned by chris
... I think we should have meaning explanations of policy assertions in the guidelines, but not in the framework
... a general statement in the framework is confusing

<Zakim> cferris, you wanted to respond to daveo's point about the rsp assertion and rm assertion

chris: about rsp / rm concerns by dave
... the language I proposed does not talk about assertions, but behaviours

<asir> +1 to Chris comment about behaviors (and not assertions

<daveroth> +1 to Chris

chris: the alternative says what something means

<GlenD> and what happens when a policy goes stale?

<cferris> how does a policy go stale?

chris: if policy provides only hints, what is the point?

<GlenD> metadata is necessarily "just a hint" and trying to imply otherwise is davegerous

<monica> See: An alternative with zero assertions indicates no behaviors. An alternative with one or more assertions indicates behaviors implied by those, and only those assertions.

dave: if you talk about behavior instead of assertions, the same concerns come up

<GlenD> it goes stale because the client pulled it over and cached it, and you've moved on and added a new optional behavior

<monica> see text from Framework

<monica> ee above

<monica> c/ee above/see above

asir: want to support chris description

<asir> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0086.html

asir: mail above shows differentiation between behavior and assertions

dave: not talking about absent assertions, but behaviors, makes everything clearer
... the "no claims" proposal: agree with chris, it leaves the door open for problems

<monica> See Section 3.2

<monica> An alternative with zero assertions indicates no behaviors. An alternative with one or more assertions indicates behaviors implied by those, and only those assertions.

monica: confused by separating behaviors and assertions
... to me this definitions conflicts with the proposal from chris

<asir> don't think anyone asked to change the definition

<Zakim> dorchard, you wanted to solicit problems with all scenarios..

monica: currently it does not get closer to resolving this issue

dave: ashok, glen and others came up with clear positions which need to be addressed to come to a WG position
... frustrated by the differentation between assertion and behavior

<monica> +1

<GlenD> assertions are statements ABOUT behaviors, no?

<GlenD> isn't it that simple?

dave: there's two positions: AIN about assertions, vs. AIN about behavior. Still confused
... need maybe another round of proposals.

paul: what are your key questions to be answered?

dave: don't know them, that is the point

frederick: thanks dave for a clear explanation
... does chris proposal mean that we don't talk not about negation anymore, or is there a different kind of negation?

<Zakim> cferris, you wanted to respond to Monica's concern regarding the differentiation of assertion type and behavior

<monica> 1+

dan: no negation tied to a negation type, it just says "I do what I say, nothing else"

<cferris> A policy assertion identifies a behavior that is a requirement or capability of a policy subject.

chris: response to monicas concerns

<GlenD> "I do what I say" necessarily means "I do what *this policy* at *this moment in time* says."

chris: definition of policy assertion meant by monica: seems to be clear to me
... I can have several assertions that can say the say thing, like "do reliable messaging"
... policy is making a statement "this is what I know".
... if we leave policy "makes no claims", what is the point of it?

dan: about the difference between behavior and assertions
... we have done ws security, rm etc.
... we have these protocols you can implement, but a requester without metadata does not know which one to use
... that is there policy and assertions come in
... you describe the behaviors using an assertions. These are separate things: the metadata and the behavior you do
... with the latest proposal from chris, you don't have to think about assertion types anymore

monica: we are talking about assertion types. We need text which makes this cleare

paul: was dan's explanation helpful?

monica: yes, but it does not resolve the issue

<cferris> in my latest proposal, there is NO MENTION of assertion types in the text I added

dale: chris and dans explanations help with what policy should do
... that is information which you should put into policy guidelines.

<fjh> explanations are helpful, thank you. Improvement not to have to look at other assertion types elsewhere.

<GlenD> +1 Dale

dale: the current proposal does not add clarification but rather confusion

ashok: question: I have one assertion which says "I do x". Can I do Y?

Dan: does x include the behavior of y?

ashok: no

Dan: so you cannot do y
... if y and x are independent, you have no idea how they interrelate

<GlenD> ...even if Y is a behavior that introduces no requirements? -1.

Dan: absence as assertion type created the confusion.

ashok: so you can do negation?

dan: yes, but it is not bound to an assertion type
... I don't need to look at all assertion types in the world

<monica> A policy alternative MAY contain multiple assertions of the same type.

monica: above is from sec. 3.2
... if we allow these type things we need to be able to differentiate
... what if there is multiple assertions? What do you do?

<Zakim> dorchard, you wanted to talk about difference between sufficent and completeness

monica: chris made a differentiation to be explicit about what it means, and we are conflicted about what it means

<GlenD> "sufficient information to obtain interoperability"!!!!!

<GlenD> +1 Dave

dave: there may be two bars about information disclosure by the provider
... listing all assertions (completeness, also optional ones) or sufficient ones for interoperability
... don't like the characterization: if you don't have AIN, you don't have interoperability
... do we need to be sufficient and complete, or just sufficient

chris: I'm confused: don't understand: what is not clear about the difference between assertion type and behavior
... think of two pointers (the assertions) pointing to the same thing (the behavior)

<GlenD> That sounds like a required behavior, Chris.

chris: we don't want it to be incomplete, it should be a tool for interop between endpoints

maryann: follow up on what chris is saying
... whould it be helpful to have a statement in the policy intersection section
... sec. 4.5, after last bullet on compatibility processing. In the lines of what dave said

<dorchard> chris, you are missing my point that there's a difference between sufficient to talk to the endpoint or completely described how can talk to the endpoint

maryann: to say that you should apply the behavior

monica: answer to chris: about the assertion vs. behavior difference

<dorchard> And I really think there is consensus that sufficient is a goal of the WG.

monica: maybe there is an opportunity to put s.t. in the intersection section (proposal by maryann)
... two different policy assertions might map back to the same assertion type

<asir> I agree with Maryann that the case explained by Monica is covered

<monica> implicit

dan: the IBM proposal has nothing about enforcement
... the policy framework can't keep you from lying, interop just suffers. Proposal does not change that
... the proposal only helps to read the behavior

dave: point I made earlier: I think there is consensus in the WG that the policy provided by a provider must be sufficient to the client
... people who want to remove AIN say the things which are optional may not be described

asir: optional is a marker not existing at the data model level
... in the data model, you have two alternatives

<paulc> 1. AIN Vocabulary flavour: Any assertion not in a vocabulary should not be applied (Original chris proposal) 2. AIN Closed favour: Any assertion not in an alternative should not be applied (revised chris proposal) 3. AIN Removal: Any assertion not in alternative means nothing. It may or may not be applied.

paul: heard no preference for 1, several for 2 or 3
... have no heard from anyone: which can people live with?

<asir> for the record the second flavor doesn't capture the revised Chris proposal

<asir> correctly

<bob> Can WS-Policy agree with our response to the WS-Policy issue?

paul: I don't think this discussion has nothing to do with the ws addressing discussion
... policy WG has never responded to the addressing WG, it's agenda item 10a

<cferris> http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0060.html

<paulc> WS-A response reply: http://lists.w3.org/Archives/Public/public-ws-policy/2007May/0060.html

paul: dan wrote a reply that he is happy with the ws addressing proposal
... bob is asking if we can reply back and say "addressing has done the right thing"?

monica: don't object, only one point about formulation related to wsp:Optional

<bob> thank you

paul: you can make a LC comment, if you are ok now with the general solution

<cferris> ACTION: Chris to respond to WS-A regarding the WS-P WG feedback on their Metadata spec [recorded in http://www.w3.org/2007/05/09-ws-policy-minutes.html#action05]

<trackbot> Created ACTION-299 - Respond to WS-A regarding the WS-P WG feedback on their Metadata spec [on Christopher Ferris - due 2007-05-16].

paul: if everybody is fine, we can go to ws addressing wg with our OK

monica: we should look at the comments from ws addressing and other domains for the framework

chris: for the record, IBM cannot live with "no claim"

paul: adjourned