See also: IRC log
<whenry> William Henry is here
<cferris> scribe: umit
Chris introduces the agenda and asks whether the minutes were reviewed by everyone from the last call.
Chris: any objection to approve and making them public?
No objection noted.
<scribe> ACTION: Felix to publish the minutes [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action01]
Asir: First action item done.
Chris: Second and third: Done.
Paul: Will do mine right now.
Chris: Last one completed.
Chris: We will go through and clarify points.
Chris reads the charter for the attendees.
Chris: This is a very aggresive schedule.
... any questions on the first two questions?
Frederick: On the first point, it is not necessary that you put all the constraints that exist in to the policy.
Ashok: There is an WS-SX issue. We would like semantics in WS-Policy that defines comparing assertions with parameters. I plan to discuss this tomorrow.
<scribe> ACTION: Ashok to desribe the issue [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action02]
Chris: This is the time to describe concrete issues and assign actions for concrete proposals. Please raise them.
Chris continues with point 3 in Scope section...
scribe: No questions on point 3.
Chris continues with policy attachment...
Ashok: What about if we wanted to attach policy to sth else such as JMS (other than WSDL and UDDI)?
Maryann: We only talk about those that are defined. This is a general mechanism.
Chris continues with Section 1.1 and 1.2...
Toufic: On Negotiation, how strongly people feel that it is out of scope?
... There is a mechanism for
intersection so it may fall into the scope.
Glen: It should not be deemed out of scope for a discussion topic. However, we may not go far to build a mechanism for this.
<maryann> toufic- the term "negotiation" being out of scope may be problematic, it is a real use case
<maryann> toufic- we have mechanisms for intersection
<maryann> if you include negotiation you need to include management ( yakov)
<maryann> (asir) when you include negotiation, its a general thing
<maryann> (asir) any negotiation is a general problem and should be solved in a general way
<maryann> (toufic) ssl has negotiation
PaulC: 2 comments. When we go farther down in the charter, we have a deliverable for charter for future work.
Please raise this as an issue. We can have a finite discussion and decide whether we want to do this for v.next.
... I don't think we should
have a high bar for our issues list. We should capture such issues in the list.
Toufic: Ok.
Chris: Mapping Policy is next...
Chris reads Alternate Representations...
Chris: Any Questions?
None noted.
Chris continues with Deliverables
scribe: Our first public draft is in July. We have editors drafts right now.
Chris reads the milestones. We have 13 months for recommendation.
No comments were made on the milestones.
Chris reviews the rest of the document and the time line.
Umit: When do we decide about the next charter split? Can we do the work, such as negotiation parallelly?
... Our charter is very constrained.
PaulC: There are several options. TF in our group or create an incubator group for additional work.
<whenry> thnx
ChrisF: As long as it does not interfere with the work of this workgroup, there are several options we can follow.
Chris continues with Dependencies.
<whenry> AER we saying that discussing SLA as part of WS-Policy are out of scope for now?
<maryann> scribe lost connection - maryann to fill in
<maryann> (chris) apologies in advance for dogs barking in the background on calls
<maryann> (chris) meetings-
<maryann> if we fall behind we may need an additional meeting
<maryann> we understand around the holidays people need more time
<maryann> communication- member list is only admin or member confidential information
<maryann> everything else should be on public list
<maryann> decisions- consensus for decisions wil be the focus
<maryann> this is not driven by roberts rule, so the chairs will seek to find consesnuse where possible, but call for decisions when the group is seen to be reaching an impass
<maryann> (paul) one technique I use, is to ask each party to represent the pros and cons of the other side of the argument
<maryann> (paul) earlier in the process that issues surface, the earlier we will reach consensus
<maryann> (paul) the chairs have the responsibility to meet the charter
<maryann> (paul) people should realize that achieving half of your issues can be seen as a mark of success for "consensus" and you always have the alternative of filing a fomal objection
<maryann> (felix) you can also come to me as the W3 rep
<maryann> (chris) one technique I use is to send people off to the bar
<maryann> (paul) there is a strong correleation between the quality of the bar and the quality of the results
<maryann> (umit) chris is going to have to take me to the bar alot
<maryann> (chris) patent policy
<maryann> any questions
<maryann> (phillipe) are people familiar with the patent policy?
<maryann> maybe felix can say a few things
<maryann> (paul) myabe a few words on when the policy kicks in
<maryann> (paul) sends mail to the AC reps, this can be an issue
<maryann> (felix) it may be that the participants get notified
<maryann> (felix) I'll take an action to find out who gets notified
<maryann> (chris) we'll take a break and make it 30 minutes and try to resolve the network connectivity issues
<bijan> So, returning at 10:50?
Chris: We remind you that we will not take a break in August.
... Let us know in advance if you are going to
take a vacation.
... We had a discussion on time slot 12:00-2:00pm EST previously. This was a difficult challenge to agree on this
timeslot.
... We considered other options, but we recommed that we leave it on this time slot until September.
Chris reviews the concall schedule in the agenda. No concall next week.
Glen: Why don't you want to have a Wed meeting next week?
PaulC/Chris: Travel takes a week. In addition, editors may need the following week for revising the docs.
Chris: Microsoft offered to host Sept 12-14 F2F meeting.
Glen: Can we have couple of slots to pick from?
PaulC: It was in the charter.
Glen: We may be able to host the November f2f meeting. I will check the availability
<scribe> ACTION: Glen to look into hosting in November... [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action03]
<uyalcina> Chris: Any volunteers for the January (15th or 22nd)? Somewhere in the WestCoast?
<uyalcina> PaulC: Dave could you look into this?
<uyalcina> ACTION: Oracle to look into hosting the f2f in January starting 15th or 22nd... [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action04]
<uyalcina> ACTION: BEA to investigate hosting F2F in January 15th [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action05]
<cferris> http://www.w3.org/2006/07/rec-track-summary.html#(3)
<uyalcina> Felix works through the slides
PaulC: How much obligation is on the wg to identify how much changed from the previous exclusion list?
...
Is the issues list sufficient diff of the html?
<ssoltysi> Hello. Seumas Soltysik from IONA. I will join con-call shortly.
Phillippe: Each step the change log is sufficient, but we can do both (including the issue list).
<scribe> ACTION: Felix to investigate the overlap case [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action06]
http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0006.html
Slide: Service Metadata
<ssoltysi> Yes.
Asir: I am just showing examples.
Maryann: Some of these examples are not in the charter? Are they Microsoft specific?
Slide 10
Glen: There is an interesting touchpoint. Policy and other metadata are related. For example, the wsa
UsingAddressing is related to this discussion.
... Such metadata is embedded in WSDL as well.
Asir: Look at the example on Page7 for using different operators are used.
Slide 11
Asir: MTOM is an example (look at Page 8)
... for an optional assertion
Slide 12
Asir: Page 9 has an example
... The dependent behaviours are represented as nested assertions.
,,, transport token and algorithm suite are examples for nesting.
scribe: Security is a complex domain that requires nesting. Complexity is hidden via nesting.
Slide 13
Asir: There are two different ways of naming. Example is on Page 10
<scribe> ACTION: Umit to file an issue on the usage of xml:id as a naming mechanism [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action07]
<ssoltysi> I am looking at the slides, but what do the page numbers refer to? are there 2 different docs that we are looking at?
<ssoltysi> When you say example on page 10, what doc is that?
PaulC: The adoption of xml:id is not present due to WS-Security impact. It is an issue to track.
Slide 14:
<prasad> What is the 2nd doc?
http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0001.html
The second document is the Understanding WS-Policy document where the examples reside.
<ssoltysi> thanks.
Asir moves on to Part 2
Asir: Page 13 contains an example.
Slide 17
Ashok: Could you motivate the normal form?
Asir: The next 2 slides will come to that
... No policy reference exists in normal form.
Page 15 example illustrates the normal form.
<maryann> (Asir) policy reference is replaced with expression
Slide 19
Asir: The Policy is not required, but it is a useful piece of metadata. The policy aware tool on the other hand
must choose one of the alternatives.
... QName of the assertion identifies the behaviour of the assertion.
... Within an assertion, the
parameters are opaque.
... there is a direct mapping between the normal form and the data model.
<bijan> +1
<bijan> er
Slide 20
<bijan> That's what I wanted :)
Asir: This illustrates the mapping
Slide 21
Look Page 21.
scribe: Contoso may be supporting several alternatives
Fabian: What if your assertions have the same attributes but they take different values?
<bijan> 1) what does "behavior of the assertion" mean
Asir: In this case, he specific domain may need to extend the compatibilty definition to extend to the parameters.
<bijan> 2) Can policies (in normal form) be incoherent? E.g., you have two equivelent collections of policy assertions
Asir: c1 and c2 are compatible with t1 and t2. (C for Contoso's Policy, T is for Tony)
JeffM: Where is the only if part?
Asir: It is in the spec, this is not complete.
Slide 22:
Asir: Example is in page 26
Slide 23:
Asir: What is extensible, what is the name and what is the processing model for extensions are the questions to
be answered.
... In this case, the existing clients do not need to change. This level of versioning support is not new.
Chris/Umit: Not everyone agrees with this definition, though.
Slide 24
Asir: The policy assertion is the key
Slide 25
Asir: There are three different assertions.
... they contain several parameters.
nested policy assertion affects the intersection algorighm, but the parameters do not.
Slide 26
Slide 27
Asir: Device profile is listed in this slide.
PaulC: How many people want to ask questions?
Many hands raised.
PaulC: Lets time bound the discussion for those questions.
Glen: What is the intent of QA? Clarification or for raising issues?
PaulC: Lets get all out on the table.
<Fabian> it's lunch break
<bijan> Ah
<bijan> But my questions are in the irc buffer
<pbc> We will be reconvening at 1:30pm CT - in about 10 min.
<abbie> ok
<maryann> Q& A session about to begin
<maryann> questions from the phone?
<maryann> first question from the archive- "behavior of the assertion"
<maryann> [asir] description of the assertion....the name of assertion indicates the behavior
<maryann> [glen] the content is the details
<maryann> [asir] parameters carry details
<maryann> [glen] tricky line between semantics of q name and semantics of the content
<maryann> [glen] careful about design of the assertion space to make sure you communicate
<maryann> [maryann]
<maryann> [paul] if this reflects to what was logged .....
<maryann> [asir] separation between qname and parameters, the qname indicates the behavior and the parameters indicate detail
<maryann> [paul] we may not have captured this comment
<maryann> [maryann] the decomposition of a domain needs to be done by experts who understand the way the framework will work in order to correctly capture the semantics
<maryann> [asir] when you have 2 alternatives, the author made duplicates, in interacting you have to chose one of them
<maryann> [paul] since policy doesn't express which option you chose
<maryann> [dale] does exactly one mean you select one of them?
<maryann> [asir] there is text in 3. something that explains it
<maryann> [dale] you select these things it is an accident if they are equivalent
<maryann> [paul] it is bad to have domains where the alternatives are orthongonal
<maryann> [ashok] this is one of the things i will bring up tomorrow
<maryann> [paul] request to discuss this further after Ashok's presentation
<maryann> [yakov] support the same issue, you can have security problmes
<maryann> because so many policies exist
<maryann> [yakov] you might need "if and only if"
<maryann> [asir] you have policies with stronger security and one with weaker it shouldn't be in the policy
<maryann> [umit] do you want a preference?
<maryann> [asir] will Ashok send slides?
<maryann> [ashok] yes late tonight ( just in time slides)
<maryann> [maryann] there are ways that the framework can enable policy domain authors to correctly represent the semantics of the domain
<maryann> [maryann] the spec does not currently have any way to indicate that alternatives are "preferred"....all alternatives are equivalent
<maryann> [paul] dale has questions,
<maryann> [dale] one of the examples ....is it required that policy use the "optional' or can they represent "optionality" through the qnames
<maryann> [dale]there is a little confusion between what part of the assertion semantics should be used by the framework or the domain
<maryann> [asir] expresses a requirement of a domain, here it is represented, best practice is to use optional
<maryann> [dale] framework provides a constraint on authors
<maryann> [asir] page 32 optional behaviors
<maryann> [dale] will best practice express these?
<maryann> [paul] the group will need to decide what is the best way to communicate
<umit> We really need a policy authors guideline document. All the issues mentioned so far are examples of the hurdles that the domain authors will face.
<maryann> [paul] we have an option to develop a primer
<maryann> [paul] having something in the normative spec might be appropriate, we have to discuss this
<maryann> [paul] we can try to make some of this more approachable
<maryann> [dale] point 2...... boundary betewen domain and framework, is there any standard device recommended?
<maryann> [phillipe] wasn't it in a previous draft?
<maryann> [asir] yes it was difficult and complex for interoperabilty
<maryann> [dan] and so it was removed
<maryann> [ashok] there is part of the framework you didn't mention
<maryann> [ashok] what are the semantics of the "absence" of an assertion?
<maryann> [asir] the absence of an assertion?
<maryann> [ashok] yes absence
<maryann> [asir] the absence of an assertion means it makes no claims about it
<maryann> [ashok] that's not what the spec says
<maryann> [ashok] if the assertion does not appear it is not ....
<maryann> [asir] this is an issue that was raised in the TX working group
<maryann> [asir] section 3.2 second paragraph
<abbie> siging off,
<maryann> vocabular is the union of all types
<maryann> [dan] so if the author has a type in one alternative, but not in another alternative, the spec is ambiguous
<maryann> [ACTION] Dan / Fabian will write up the issue for 3.2
<maryann> [ashok] so what if the type is not part of the vocabulary
<maryann> [umit] what is the vocabulary for the web service can include other information not expressed by policy
<maryann> [paul] it is possible that someone could construct a message that succeeds in being "accepted" by the web service
<maryann> [maryann] the spec and the framework are meant to be declarative
<maryann> [umit] the problem is the actual statement ( which is the subject of the action for Dan & Fabian)
<maryann> [asir] from the framework level there is one intersection model
<maryann> [asir] domain authors can extend this
<maryann> [asir] there are delegation models for implementation
<maryann> [glen] there is an issue here
<maryann> [glen] one of coverage....being complete
<maryann> [glen] if the framework had a "must understand" bit it indicates that you know what to do when you see that qname, without that bit, you might get the right results
<maryann> [paul] can i help? this is like a subtyping hierarchy
<maryann> [paul] intersection model tells you how to do top type comparison
<maryann> if you have canadian addresses and us addresses, how do declare what the comparison deals with, does the subtype have to know how to compare fields in the canadian address
<maryann> [paul] will someone own writing up this issue?
<maryann> [paul] if you have optional content defined by the domain assertion how do you indicate to the policy engine doing the intersection when or if it should take that content into consideration
<maryann> will someone own this issue and track it on the mailing list
<maryann> [glen] is there a formal way to track issues
<maryann> [paul] yes this is on the homepage
<maryann> [ruchith] nested policies, can be used as a way to constrain the subtypes
<maryann> [glen] there are the same limitations on the nested assertions
<maryann> [ruchith] there are examples in security policy
<maryann> [paul] yes these will be covered when Tony does a presentation tomorrow on security policy
<maryann> [dale] there might be some standard way to do this,
<maryann> [paul] ashok is familiar with this topic as "deep equals"
<maryann> [paul] spent 5 years trying to get agreement on this
<maryann> [paul] its very comples
<maryann> [dale] is that needed here
<maryann> [paul] don't know
<maryann> [ashok] sometimes people want exact match, there are many variations
<maryann> [paul] you can do this in sql
<maryann> [seamus] general question- processing model thoughts, recursive, can result in complex processing, if you look at the apache framework is to create a wrapper around the DOM model .....processing assertions this
<maryann> [maryann] there is some text warning about that in 3.1 the last paragraph
<maryann> [paul] what has apache done?
<maryann> [they have represented an assertion as a wrapper around the
<maryann> DOM you have to walk the DOM tree, and they don't have a framework
<maryann> they are simply giving you a DOM
<maryann> [ruchth]
<ssoltysi> http://ws.apache.org/commons/policy/index.html
<maryann> there has been a problem identitfied .... i can talk about implementation
<maryann> and there is new code checked in
<maryann> there is an assertion added
<maryann> domain specific authors decide
<maryann> [pau] the first implementation does the api level
<maryann> [paul
<maryann> the second approach is to use domain assertions
<whenry> quit
<maryann> [maryann] do we discuss implementations here?
<maryann> [paul] as to Maryann's point, we probably need to do some due dilligence to look at this and see what./if we need to add text
<maryann> [ashok] client is doing matching and server is doing matching...they better be doing the same domain specific semantics or you may get different results
<maryann> [pau] this is part of XML processing
<pbc1> Domain authors should be cognizant of the processing requirements when defining complex assertions containing additional assertion content or nested policy expressions. Specifically, domain authors are encouraged to consider when the identity of the root Element Information Item alone is enough to convey the requirement (capability).
<maryann> [paul] this is an important piece of information for developers
<maryann> question to Seumas--- will you open an issue for this text?
<maryann> ACTION to Seumas to open an issue
<umit> Seumas to open an issue
<umit> ACTION: Seumas to open an issue [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action08]
<asir> Scribe: Asir S Vedamuthu
<umit> ACTION: Dale/Fabian to write up the issue for Section 3.2 [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action09]
[Presenter: Maryann Hondo]
Slide: Policy Attachment Technology
Glen: what does it mean to attach a policy to a specific policy subject?
Ashok: Oracle has created domain expressions that apply to other domains (JMS ..)
Paul: are you asking for an agenda item in this F2F or future meeting?
Ashok: future
<scribe> ACTION: Ashok to raise an agenda item on domain expressions that apply to non Web service constructs [recorded in http://www.w3.org/2006/07/11-ws-policy-minutes.html#action10]
Frederick: why the RECOMMENDED means of associating a Policy with a Policy Subject that has a WSDL 1.1 description is to attach a reference ...
Slide deck - http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0012.html
Toufic: key references - are there any way to categorize this using the tModels?
Maryann: this is a proposal and needs to engage the UDDI experts (address issues, if any)
Toufic: this is a concern in the UDDI attachment mechanism, will file issues as appropriate
Maryann: explore interpretations on how WSDLs are expressed in UDDI
Paul: you mentioned few issues. Do you plan to point them out as we walk through the specs?
Maryann: yes
Ashok: request to elaborate on effective policies for WSDL
Slide: Example- stock quote
Slides 10 and 11
Glen: request to clarify the partitioning of WSDL constructs into policy subjects and the semantics of associating policies to policy subjects
Umit: need clarification on the granularity of attachment points for a given assertion
<cferris> glen: think that this is not about any given domain, use the domain asertions as examples, but talk about wsdl as a description and say that there are a generic set of rules for endpoint, port, binding, etc.
<cferris> mah: so is this a wsdl thing or a policy thing?
<fsasaki> Paulc: You say there is not enough in attachment about how to handle that?
<fsasaki> Glen: there is enough, but brief
<fsasaki> .. now we need to say: this needs to get right
<fsasaki> no coffe break
<fsasaki> paul explains http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0007.html
<fsasaki> PaulC: if you raise issues like "TBD", they go down the list
<fsasaki> .. at least describe the scope of your issue
<fsasaki> Plh: I had people who sent issues without test cases. Later it's the editor's have to write test cases.
<fsasak> PaulC: the example given on "namespace URI" in the mail is just an example
<fsasak> example issue is at http://lists.w3.org/Archives/Public/public-ws-policy/2006Jul/0008.html
<fsasak> PaulC: I have a very low bar for issues
<fsasak> .. with bugzilla, everybody with an bugzilla account can create an issue
<fsasak> .. Chris has volunteered to use an XML based issues list for now
<fsasak> .. if you please use the template given in the mails above
<fsasak> Glen: using bugzilla or other (automatic) tracking systems might help
<fsasak> chris: We had some contacts, but don't have a reply yet. XML-P is using the XML-based format
<fsasak> .. as we get the other system, we can change (in this early change)
<fsasak> Ashok: I strongly recommend bugzilla, the W3C public version of it
<fsasak> http://www.w3.org/Bugs/Public/
<fsasak> paul shows how to enter a new bug for XQuery specs at http://www.w3.org/Bugs/Public/enter_bug.cgi?product=XPath%20%2F%20XQuery%20%2F%20XSLT
<fsasak> Paul: problem of bugzilla: everybody with a bugzilla account can reply to issues
<fsasak> .. sometimes it is hard to tell whether people debate on bugzilla or on the mail list
<fsasak> .. many WG let bugzilla sent mails to public lists
<fsasak> .. the discussion is on the public list, and you manually update the entry after a decision
<fsasak> .. that is very important, because later in the W3C process you need to make explicit who (dis)agreed with your changes
<fsasak> Maryann: somebody enters an issue, and the public list gets a notification
<fsasak> .. who is responsible to make that entry / and who is updating bugzilla after decisions?
<fsasak> Paul: the people responsible for an issue. Esp. at last call, you put templates into bugzilla entries
<fsasak> .. saying "if we don't hear from you, we will close the issue in two weeks"
<fsasak> Umit: the issue list in WSDL was very efficient
<fsasak> .. the whole history, e.g. how you reached the conclusion etc. , was much more comprehensive
<fsasak> Jeff: is a kind of taste question
<fsasak> .. I have been in groups who use both. I like the issues list. It is only difficult for people who have to maintain it.
<fsasak> Paul: the advantage of bugzilla is that you can distribute the work
<fsasak> .. esp. at LC, that can be very helpful
<fsasak> Ashok: in Bugzilla, you can add the version you are looking for
<fsasak> Paul demonstrates how to search existing bugs in Bugzilla
<Ashok> I meant you can carry on the discussion of the bug in Bugzilla itself
<fsasak> Umit: if you use the tool, we should have guidelines how to record decisions etc.
<fsasak> .. using the tool by itself will not be very effective
<fsasak> Paul: there is tutorial material from XQuery about this
<fsasak> Paul: via bugzilla, we distributed the work to the public
<fsasak> .. who had to use bugzilla
<fsasak> .. in our agenda, we pointed to the bugzilla entries
<fsasak> .. we started discussions on the (member)list until we get consensus
<fsasak> .. then somebody took an action to make the update in bugzilla
<fsasak> Chris: we also just used a link to the minutes
<fsasak> Paul: I am neutral here
<fsasak> .. I don't assume more than 100 comments
<fsasak> .. I am o.k. with using the XML version
<fsasak> .. it will take some time for the editors to participate in this
<fsasak> Chris: we have two weeks to decide this
<fsasak> .. until the next distributed call we will check bugzilla, and tracker
<fsasak> Paul: tracker can also track action items
<fsasak> .. so we want to make a decision until the 20th
<fsasak> bugzilla description for QT is at http://www.w3.org/XML/2005/04/qt-bugzilla.html
<fsasak> Paul: we have a conflict tomorrow morning, so we might change plans
<fsasak> .. we will convene this meeting right at 10 a.m.
<fsasak> meeting adjourned, this meeting will start 10 a.m. tomorrow morning