WS-Policy F2F Meeting -- 13 Sep 2006

agenda item13 "tarball" on extensibility

paul: update on primer
... agreed to 2 documents
... this is important because one or more of your items may need to be targetted to one or more of the documents

david: there is a missing step to publishing the primer

paul: the assumption is that the editors will convert the understanding paper into the prime

primer

paul: do it on reqtrac with the flexibility of publishing later
... modified the agenda, the first item this morning is item 13

<asir> Agenda: http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0041.html

<toufic> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0041.html

paul: bugs 3617, 3590, 3662
... digression- who has bugs outstanding

glen has one

toufic has 2

paul: we currently have 30 bugs

daveorchard: is success measured by moving the number up or down?
... tie in the versioning info for the primer

asir: item 12 on the agenda

daveorchard: we need to discuss this

paul: action #28 is being moved to be part of the dicussion of item c under agenda topic 13

versioning policy is not clear

<asir> Most recent e-mail on 3617 is http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0045.html

<Fabian> got it

paul: looking at bug 3617 and message thread
... asir, can you summarize umit's proposal

asir: namespaces in policy framework
... 3rd bullet in section 2.3

<PaulC> Modifications to the pattern facet of a type definition for which the value-space of the previous definition remains valid or for which the value-space of the preponderance of instance would remain valid.

paul: the material after the or is the same as the information before the or

asir: doesn't hurt to change the 3rd bullet
... formulated a concrete proposal

dave: the proposal for the 3rd bullet, would have expected you keep the other part

asir: read the 3rd bullet as dealing only with pattern facet

jeff: it doesn't say that though

paul: the second clause doesn't relate to the pattern facet,

dave: there are many topics here

jeff: what does preponderance mean?
... i've only seen it used in a legal context
... the majority of the instance ?

fred: clear definitions might be put in the primer

dave: there might be a case where some cases might not be valid as we go forward
... backward compatible not by type but by usage instances

jeff: so someone is considering a change, what's the test to see if it meets this criteria?

dave: you go to the working group and it comes up with its own hueristic

fred: why preponderance?

paul: to announce that the working group will make this decision

jeff: if we didn't have this verbiage, if we made any changes we'd have to have a new namespace

paul: no

pau: we have a difference of opinion

paul: david has one interpretation,

dave: reconsidered, and i agree with asir
... this bullet has raised a good point and i don't want this point closed off
... move the preponderance of instances to apply more broadly

paul: look at text before the bullets
... this is just examples not all cases
... what do you want to do to bullet 3?

dave: trying to figure out what umit was referring to, i prefer to keep the status quo

paul: why isn't it a repetition?

dave: its talking about a pattern facet changing, and that some of these are not valid

paul: leave the text and explain to umit, the first part has no impact, all the old cases remain valid
... the second part says if we change the pattern facet does not impact the majority of cases, the working group can agree to not change the namespace
... its common for the working group to lock down the namespace in CR

jeff: preponderance means 50% +1
... vast majority is different than preponderance

fred: i don't understand either
... what are we saying? its majority vote?

paul: its always consensus
... we're leaving it subjective

jeff: there's still an issue, if i have an early implementation, what does my implementation do with this thing it doesn't expect

paul: summarize..... we change "preponderance" to "vast majority, no change to bullet 1 and make the change to bullet 3

fred: what's the 4th bullet?

david: you can change value by changing min occurs or extending max occurs

ashok: its peoples inability to parse the sentence

<dorchard> technology is hard. c'est la vie.

<PaulC> Re bullet 1: No change is necessary

<asir> Related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/23

<PaulC> Re bullet 2: change "preponderence" to "vast majority"

<PaulC> Re bullet 3: s/cardinality of elements/cardinality of elements (i.e.

<PaulC> modifications to minOccurs or maxOccurs attribute value of an element

<PaulC> declaration)/.

paul: we need someone to update the bug with this decision

asir: combine 3590 and 3662

paul: why do we have 2 bugs?

dave: because you told me to

paul: can we mark 3590 as a dup?

dave: 3590 the key thing is to have an element extensiblity point for reference

<dorchard> 4. The PolicyReference Element is modified to add an element extensibility

<dorchard> point. This should be for any namespace, which means a slight change to the

<dorchard> notation section. This includes specifying that unknown element child content

<dorchard> is ignored.

dave: propose for any namespace, asir prefers other

<PaulC> Point 4 in 3590:

<PaulC> 4. The PolicyReference Element is modified to add an element extensibility point. This should be for any namespace, which means a slight change to the notation section. This includes specifying that unknown element child content is ignored.

dave: add element extensibility and close the issue
... then we can decide in 3662 which one

RESOLUTION: for 3590

<asir> Related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/23

RESOLUTION: for 3590 is adopting the first sentence in point 4 - The policy reference element is modified to add an element extensibility point

<dorchard> #4 is agreed to by WG, but element extensibility of ##other vs ##any is decided in 3662

RESOLUTION: for 3617 - make changes to bullet 1,2 3 as indicated above and anchored by http://www.w3.org/@

http://www.w3.org/2006/09/13-ws-policy-irc#T16-41-41

bug 3662 and agenda item 13 c

paul: how many people think it should be ##any, ##other,no opinion
... ##any ? - 6
... ##other? 2

RESOLUTION: 3662 is ##any

TOP

<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/25

agenda item 12

<PaulC> Re 3662 no one could not live with ##any

paul: this proposal is changes for primer

dave: guidelines vs primer

paul: primer is aimed at people to understand the framework and the guidelines are for policy authors where does this belong?

dave: this is a little fuzzy
... the primer is for people who don't understand the spec
... and guidelines is for people who are writing assertions
... there is a secition in what is now the "primer" that talks about versioning
... so

paul: so will you update the section?

dave: augment

asir: this is extending the language

<asir> CVS version of the primer is http://dev.w3.org/cvsweb/~checkout~/2006/ws/policy/ws-policy-primer.html?content-type=text/html;charset=utf-8

paul: what's wrong with targetting this at the primer?
... and then seeing what if anything should move to the guidelines?

dave: yes

paul: is your material to supplement or replace section 3.7

dave: there's also 4.4.7
... this would be a new 4.4.8

asir: 2 questions- as a user of the framework how do i extend the framework
... how can 3rd parties extend
... what is the working group strategy for v.next
... separate the content to determine if this goes into the nornative document

dave: the problem is that....i gave a number of scenarios, it gives suggestions, but that's the extent
... there's a couple of best practices
... the material in here is not normative, and not worth being promoted
... its more examples

asir: i will point out..... " we can imagine a future version"

paul: is there any material that belongs in the framework doc in a non-normative annex

vlad: need to consider assertion version in the guidelines

RESOLUTION: accept this for text to the primer

<prasad> http://lists.w3.org/Archives/Public/public-ws-policy/2006Aug/0159.html

s/this/http://lists.w3.org/Archives/Public/public-ws-policy/2006Aug/0159.html

<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/26

Clarify the relation of overlapping definitions in the fr...,

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3562

RESOLUTION: fixed with the working group draft -- one of the editors to update bug

3559 - Conformance Sections needed for both specs

<PaulC> proposal in http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0027.html

proposal: http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0027.html

<Fabian> yes, can see

paul: does this address 2.0 proposal
... if we adopt this there may be an issue

dave: the value facet might change

paul: for what data type?

dave: you might find an issue about conformance....with the preponderance

paul: we would have changed the schema without changing the namespace
... if there were complaints we wouldn't do that

monica: i sent a note to umit, do we want more detail with respect to conformance
... there is work in
... OASIS on conformance
... asking if we want more detail

paul: this is the same level as wsdl

monica: yes, but do we want more detail

<monica> See: http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=ioc

paul: well here's a proposal to put it it
... there would be a way by putting more MUSTS in the spec

monica: if you look at the reference above, you can see the difference between basic and advanced functionality

paul: exit criteria for CR will tease this out

monica: we just might want to address some advanced functionality

<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/27

RESOLUTION: 3559 with the text in http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0027.html

issue 3705

<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3705

http://www.w3.org/2005/06/tracker/wspolicy/actions/48 will be an action for the next meeting

break

paul: startup again
... glen opened bug 3720 --- terms should be defined....

<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3720

<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/28

paul: there is a new bug 3720
... not sure how to deal with this yet

17. New issues

a) Using UsingAddressing Extension Element as a WS-Policy assertion, PLH

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3656

asir: in the addressing specification there is a "UsingAddressing" element
... which can be used as a wsdl element or a policy element
... question from phillipe on whether the current doc should have the "generic" text replaced with a more specific reference to our spec

http://www.w3.org/TR/2006/CR-ws-addr-wsdl-20060529/#id2263339

ashok: simple question...says points to a policy assertion.....is this one assertion?

glen: the text was meant to be generic since the policy framework was not yet within the w3c
... question is, should we now make it a wsp:Policy assertion

asir: i can't see why there wouldn't be a normative reference
... there are 17 references already in the primer
... timing will be a challenge
... phillipe looked at the examples in the primer and approved

paul: this is a process heavy issue---- was raised at the CG

fabian: i have some confused questions about who is defining the assertions

<paulc> glenn: +1

fabian: we need to make clear that its addressing that should be defining the assertion but its ok to have a reference in the primer

paul: are you talking about 3619?

fabian: no

<GlenD> (I was +1-ing the idea that we wouldn't talk about the assertion, and leave it to WSA to define it and any text about how the WSDL extension and the Policy Assertion would coordinate)

vlad: more common question....if addressing defines this, its "outside" the framework if its a wsdl element
... it says it "might be an assertion"

asir: it can be either a plain wsdl assertion or a policy assertion

vlad: how do we process it if its a wsdl element?

asir: that's the 3rd question

vlad: its about all kinds of extensions and how they relate to alternatives

paul: this may be related to the other thread

paulc: we have to gain some experience with this

jeff: should we separate this into 3 items/

glen: great for wsa to put in a reference to policy and they should define what the assertion is and what it should mean if there are the two

paul: that's the ideal, but you ignored the process problem, because they would have to go back to working draft and they are already in cr
... maybe the text in the wsa doc should be something like "as a policy assertion in a policy framework" ---such as wsp: Policy

jeff: we don't want to restrict the e.g. to just wsp:Policy

<paulc> From WS-A WSDL binding:

<paulc> (e.g., as a policy assertion in a policy framework

<paulc> maybe this could changed to:

<paulc> (e.g., as a policy assertion in a policy framework such as WS-Policy

jeff: the group could do this as a separate note
... you could do this as a req track note

paul: and it would normatively tie the two together
... suggest that if the timing of a normative change would impact their work, they could do something like the "such as"

<asir> See http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0023.html

paul: you didn't answer the question, you ack'd the issue
... what happens if they're both there?
... if you want to make the information more broadly understood, you would express it in both ways
... suggest that at a minimum we should suggest they make the non-normative reference

prasad: as long as they are logically consistent, you could use both representations

RESOLUTION- 3656 is to make the non-normative change as a minimum, #2 yes there are examples in our primer and #3 we don't think there's an issue with both methods of expression

<scribe> ACTION: Asir to respond to phillipe and bob with the resolution for 3656 [recorded in http://www.w3.org/2006/09/13-ws-policy-minutes.html#action03]

<trackbot> Created ACTION-104 - Respond to phillipe and bob with the resolution for 3656 [on Asir Vedamuthu - due 2006-09-20].

<paulc> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3672

Clarify the policy model for Web Services

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3672

yakov: expand the scope of the usecases
... includes 5 entities in a web services model
... each of these might specify policies
... example given was authorization
... don't see any limitations in the specification to preclude this
... replace requestor/provider with entity

asir: do we get time to review this?

paul: how long do you need?

i asserted that Tony had some concerns and might not have time to repsond

paul: put the discussion on the agenda for tomorrow morning

a) Need a URI structure to refer to WSDL 1.0 definitions, etc. , Ashok

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3599

<paulc> Proposal in: http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/att-0033/Proposal_for_Bug_3599.pdf

ashok: section 4 allows you attach policies to wsdl 1.1 so this indicates the spec sees attachment to 1.1 important but when you try to do this with external attachment you have no uri mechanism
... you may not be able to add the policies to the wsdl
... this spells out a uri scheme to refer to wsdl 1.1 components using the external attachment
... take the uri and use it as a domain expression
... use the algorithms in 4.1 to work out the effective policies
... at the end also talk about how to attach to other things
... talk briefly about http and jms

paul: obvious question, wouldn't other people find these uri's useful
... is there a model of where this goes?

ashok: that's one of the things we can talk about
... can't ask wsdl because they're not going to work on 1.1 anymore
... whether we want it as normative or non-normative thats ok

paul: can it be a note?

ashok: that is another option

jeff: notes you don't maintain

ashok: what about as an appendix?

vlad: does anyone own 1.1?
... don't think policy should really own this?
... what about wsi?

jeff: too difficult to do that

prasad: clarification.....we refer to wsdl 1.1 or 1.0?

ashok: typo

prasad: when you have your namespace fragment your token should be 1.1

ashok: yes that's a useful thing

<prasad> ac pr

asir: question..in attachment draft there is a section 4 that talks about attaching policy to wsdl 1.1 ... this describes a model and a mechanism....is any of this material required to make this work?

ashok: this is something different
... using the external attachment are different ways of doing the same thing.....to do when you can't annotate the wsdl

prasad: here you inline it so not required for section4

asir: uri fragment identifiers are pieces of domain expressions
... domain expression is independent...considering the timing an the amount of work to make this work why is this in scope?

ashok: what work?

asir: more work needs to be done to make this into working draft

paul: ashok is asserting that this can be dropped in....
... there might be a sentence or two that needs to be added

asir: why are we considering domain expressions as part of the work of the working group

paul: not sure this is domain specific

dan: this is domain dependent

paul: why does the external exist then?

asir: external attachment has points of extensibility section 3.4

prasad: so this is within the charter because its just defining a wsdl 1.1 external attachment

jeff: are you making an explicit request for this to be out of scope?

asir: phillipe has mentioned that the w3c sees this as out of scope

vlad: then we make external attachment unusable for implementors

paul: there are no motions on the floor
... there are no examples
... we need an example of how it would be used

ashok: you would put this in the external attachment where it says domain expression
... need to look at the schema for applies to
... so I'll have to wrap it in an element

paul: if you have a definition of an element it would be a usage of the extensibility point

dan: i see that we're supposed to create an attachment for 1.1 and trying to
... to do another one, but is it possible to copy the wsdl and inline it using the existing mechanism
... now we'll have 2 mechanisms

ashok: wsdl 2.0 spec defines this and you can do exactly this with 2.0

paul: if you put this is the spec, we will need to test this in CR

dan: the current proposal for 2.0 is an inline method only

paul: same issue exists for 2.0

vlad: this could be an extension to the existing document for only wsdl 1.1

prasad: bigger question on the table is are we in scope or not?

ashok: maybe i should talk to my lawyer :-)

paul: we have an idea of what the technical material is
... applies to requires element content so anything inserted has to be wrapped
... ms and w3c assert that this is out of scope
... oracle, sap, layer 7

yakov: include it

<prasad> From the charter it seems out of scope but could be a useful (simple) addition

paul: dale "general purpose" mechanism .... ambiguous
... glen, monica, its useful

fred: how much work?

asir: should we ask who will implement?

paul: too early to ask this
... at risk...when you are in CR you can delete something if you didn't get successful interop, you have to go back to working draft, so when we go into CR the question will be asked
... you explicitly identify features at risk when you go into CR

touf: how is the decision made to remove it?

monica: doesn't this relate to my question on exit criteria?

paul: no

monica: you can't use exit criteria to indicate something is optional?

paul: just because something is optional doesn't mean you can get away without testing it

jeff: proper way to state it, is later on in the process we will set what the exit criteria is for CR
... to some extent we have latitude to set that

paul: the minimum is 2 and w3c prefers one open source

jeff: some set of things you have 4 implementation, for others you might only have pairwise interoperabiltiy

<vladB> q

paul: we have to tell the director when we go into cr

fred: what is the risk?

<Fabian> can we please honor the queue?

fred: in scope

fabian: i don't think anyone can interoperate on external attachment
... as it is today

paul: i have always thought we had to pick a domain for testing

fabian: this is not the same thing
... i don't see why we couldn't define the same thing for policy attachment
... in scope

jeff: its premature to talk about what is at risk
... not the concept of at risk, but the specific features that are at risk

<Fabian> I was saying: Internal attachment to WSDL 1.1 is clearly defined, no reason why we couldn't do the same for external attachment

asir: more technical work to be done....

paul: we have a sketch

asir: there's is an rfc that talks about fragment ids, for 2.0 there was also a new media type that needed to be defined

paul: the spec didn't define a media type?

asir: 3023 is the rfc

paul: question is ...are there any conflicts with these frag ids and the way people do it today>

ashok: we will look at that

paul: so unless we define a mime type and the frag ids within the mimetype we might have an issue

<asir> Paul - please project http://www.w3.org/TR/2006/CR-wsdl20-20060327/#ietf-draft

paul: so if we have these fragment ids we need an annex in the document to define the mime types
... this could be why the w3c thnks this is out of scope
... summary- the current proposal needs some more work for sectoin 3.4 and there is a concern for defining fragment ids

<scribe> ACTION: paul to discuss the summary of the proposal with the w3c [recorded in http://www.w3.org/2006/09/13-ws-policy-minutes.html#action04]

<trackbot> Created ACTION-105 - Discuss the summary of the proposal with the w3c [on Paul Cotton - due 2006-09-20].

<Ashok> scribe: Ashok

Restarting after lunch

<vlad> quit

PaulC: Consider using XPtr to point directly into the XML --- Bug 3599

<scribe> ACTION: Ashok to review 3599 proposal ked using XPtr [recorded in http://www.w3.org/2006/09/13-ws-policy-minutes.html#action05]

<trackbot> Created ACTION-106 - Review 3599 proposal ked using XPtr [on Ashok Malhotra - due 2006-09-20].

Items 18b and c queued up for Thu morning

Agenda for this afternoon:

20. Issues requiring more discussion or proposal (con't), Chair (11:00 am PDT)

d) Optional Assertions may not be usable in all circumstances, Umit

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3564

e) Semantics of successful intersection determined by domain-specific assertion content, Glen D

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3577

<scribe> ACTION: item to [recorded in http://www.w3.org/2006/09/13-ws-policy-minutes.html#action06]

http://www.w3.org/2005/06/tracker/wspolicy/actions/83

f) The absence of an assertion should not mean that the behavior is "explicitly prohibited"

http://www.w3.org/Bugs/Public/show_bug.cgi?id=3602

WS-Policy F2F Meeting

13 Sep 2006

Attendees

Contents