Web Services Addressing Face-to-Face

7 Nov 2005


See also: IRC log


Abbie Barbir (Nortel Networks)
Andreas Bjärlestam (ERICSSON)
Glen Daniels (Sonic Software)
Vikas Deolaliker (Sonoa Systems, Inc.)
Paul Downey (BT)
Robert Freund (Hitachi, Ltd.)
Hugo Haas (W3C)
Marc Hadley (Sun Microsystems, Inc.)
David Hull (TIBCO Software, Inc.)
Yin-Leng Husband (HP)
Amelia Lewis (TIBCO Software, Inc.)
Mark Little (Arjuna Technologies Ltd.)
Anish Karmarkar (Oracle Corporation)
Paul Knight (Nortel Networks)
Jonathan Marsh (Microsoft Corporation)
Jeff Mischkinsky (Oracle Corporation)
Nilo Mitra (ERICSSON)
Eisaku Nishiyama (Hitachi, Ltd.)
David Orchard (BEA Systems, Inc.)
Gilbert Pilz (BEA Systems, Inc.)
Tony Rogers (Computer Associates)
Tom Rutt (Fujitsu Limited)
Katy Warr (IBM Corporation)
Ümit Yalçınalp (SAP AG)
Prasad Yendluri (webMethods, Inc.)
Dave Chappell (Sonic Software)
Francisco Curbera (IBM Corporation)
Jacques Durand (Fujitsu Limited)
Michael Eder (Nokia)
Marc Goodner (Microsoft Corporation)
Arun Gupta (Sun Microsystems, Inc.)
David Illsley (IBM Corporation)
Philippe Le Hégaret (W3C)
Ales Novy (Systinet Inc.)
Mark Peel (Novell, Inc.)
Rich Salz (DataPower Technology, Inc.)
Jiri Tejkl (Systinet Inc.)
Mike Vernal (Microsoft Corporation)
Steve Vinoski (IONA Technologies, Inc.)
Steve Winkler (SAP AG)
Rebecca Bergersen (IONA Technologies, Inc.)
Mark Nottingham
Tony Rogers, Yin-Leng Husband, Katy Warr, Anish Karmarkar, Jeff Mischkinsky


<TonyR> scribe: tonyr

Action Item Review

Working draft Issues

Clarifying i56/i57

RESOLUTION: i56 resolved as MUST, not SHOULD

Wordsmithing the contents of 4.3 in WSDL document

disagreement on whether "include" makes order unimportant, or not

Suggestion that we may require a new CR issue regarding order

DaveH: will raise CR issue relating to this item

Revised text reads:

when present, the value of the [reference parameters] message addressing property for a message sent to an endpoint MUST

include each child element of the wsa:ReferenceParameters WSDL extension element.

RESOLUTION: i57 resolved with the revised text above.

i65 what to do when SOAPAction and default action pattern conflict

Katy explains the situation

discussion of proposal 1

MarcH: do we think there are two separate action values, or is there one?

Jonathan: we decided there should be one

DaveH: are there potential issues when using an SMTP binding?

Jonathan: are you saying that proposal 1 is HTTP/SOAP centric?

discussion circling that question

DaveH: can we simply say "prefer the action in the binding" vs "prefer the action on the wire"?

Anish: does this only apply with UsingAddressing is present?
... if so, then WSDL with UsingAddressing is new, and should not contain conflicts

Others: but should be able to migrate WSDL just be adding the UsingAddressing marker

Anish: should fix up the WSDL when adding the UsingAddressing marker

MarcH: yes, it's a simple change, but requiring lots of people to change is not good

Gil: doing it Anish's way raises the barrier to migration to WSA for the masses - the only people who care are in this room

Jonathan: maybe we should let SOAPAction die out?

Glen: but sometimes we want the action outside the XML

<yinleng> Jonathan: Logical conclusion is loosening the requirement a little. To accommodate the goals, is to relax the restriction

Jonathan: introducing Proposal 3 - loosening the requirement that SOAPAction must be identical to [action]

PaulD: I have thousands of WSDLs - how much change must I make to them?
... not at all keen to hand-edit all of them

discussion of the implications to hand-editing - suggestion that tooling should fix these problems

PaulD: some web services implemented in Perl despatch on SOAPAction

<yinleng> Katy: I don't see it as a major hack

Jonathan: several ways to solve the problem
... 1. stop using SOAPAction

Hugo: want to speak against Proposal 3, for Proposal 1
... the definitions of the two seem to be the same

<yinleng> Hugo: This is about using an URI to identify the message

Gil: problem: SOAP 1.1 doesn't have defaulting rules, unlike WS:A
... where's the problem?

<yinleng> Jonathan: Nice to have a simple method to set the action once

Glen: we have all these rules that specify what to use for action in the absence of wsaw:Action
... if there is a wsaw:Action, then of course we use that

<yinleng> Anish: It is really a WSDL extension, a binding specific thing, a binding level extension

<scribe> Chair: let's drop Proposal 3

Proposal 1: 11

Proposal 2: 3

<vikas> proposal 1

<prasad> Proposal 1

<scribe> Chair: any objections to closing i65 with Prop 1?

<yinleng> No objection to proposal 1 being adopted

RESOLUTION: close i65 with Proposal 1

i59 - async

<yinleng> Break at 10:30

PLAN: break at 10:30, extended break to resolve network issues


CR9 relates to the Action for undefined faults

<yinleng> Hugo: Mark did a friendly amendment

Hugo: explains his concern about the use of MAY twice

Hugo dictates his suggested revision.

Jonathan: not keen on phrasing that suggests that all generic SOAP Faults must be defined in the WSDL - trying the discourage that in WSDL 2

MarcH: trying to state: use this Action for all addressing faults, for example

<yinleng> Hugo: When the value of the action property has not been defined explicitly, (for example, by SOAP Modules or by an extension such as the WS-Addressing WSDL Binding) the following value MAY be used:

Jonathan: this is intended to be used for generic faults, things one would not include in WSDL normally

<yinleng> Hugo: A bit confused about Section 5 of the SOAP Binding.

Hugo: willing to drop my version

<scribe> Chair: so we are left with Jonathan's proposal as amended by Marc H.

Jonathan: this is to be inserted in Section 5 of the SOAP Binding

Umit: why is a generic fault becoming an addressing fault?

Marc/Jonathan: no, this is the required action (required because we are using Addressing)

<yinleng> MarkN: Any objections to this as resolution to CR9?

<Tony> scribe: Tony

RESOLUTION: close CR9 with Jonathan's proposal as amended by Marc

Break now for 20 minutes - back 10:55 local time

We're back


Anish explains the issue

<yinleng> Tony: proposing wording already inserted in section4.2, SOA Binding

Anish: change SOAPAction HTTP header to SOAPAction HTTP Request header (two places), and change wsa:Action header to [action] property

<yinleng> Mark: Any objection to using this as resolution to CR11?

RESOLUTION: close CR11 with the revised text as above

CR12 (DaveH's new issue)

DaveH describes the issue: question of whether the reference parameters are a list (order important) or set (unordered)

<scribe> Chair: so you are requested a statement in the Core stating that the reference parameters are unordered

DaveH: yes

<yinleng> Jonathan: Do we have to specify this? In what way will interoperability be affected?

<scribe> Chair: looking at the text in the Core document

GlenD: consider another spec which wants to sign the serialisation of the reference parameters - it would not work interoperably

<yinleng> Tony: We want to specify explicitly that it is not ordered

DaveH: suggestion - change opening statement in 2.1 under [reference parameters] from "a number of" to "an unordered collection of"
... [metadata] should change too

General wordsmithing

RESOLUTION: the WG clarified that the occurrences are unordered, and the editors will construct appropriate language to reflect this.

CR10 - introduction

Chair describes the issue

<scribe> Chair: DaveO will be supplying a proposal, but it does not appear on the list

General discussion of the subtle change that was proposed during the last meeting

Jonathan: not keen on the strength of the language in the existing words.

MarcH: perhaps replace "is contrary to Web Architecture" with "is beyond the scope of Web Architecture"

TonyR: perhaps "is outside the scope"

General wordsmithing ensues

<vikas> Perhaps change "violote" to "does not conform" ?

<mnot> proposal:

<mnot> In the Web Architecture, resources are identified with URIs. EPRs whose properties other than [address] are used to identify resources which operate outside the scope of the Web Architecture, which may result in the loss of some of its benefits. When building such systems, care should be taken to weigh the trade-offs inherent in deploying resources that lose some benefits of the Web Architecture.

<vikas> But Mark they are in scope they just don't conform

<prasad> Other properties qualify the URI (address) that really identifies the resource?

Interchange of opinions faster than the scribe can capture them...

Chair makes a new proposal

<mnot> In the Web Architecture, resources are identified with URIs, which brings certain architectural benefits. EPRs whose properties other than [address] are used to identify resources lose some of those benefits. Therefore, when building such systems, care should be taken to weigh the benefits brought by non-URI identifiers against those brought by the Web Architecture.

Nilo: properties other than address aren't used to identify

Glen: a URL can be used in two ways - to identify, or not

<vikas> Are people deliberately eschewing the word "state". Cookies bring state so does wsa:ref

<scribe> Chair: seeing it from TAG's viewpoint, important to ensure that people realise that they "shouldn't do that"

<prasad> Isn't having operations other than HTTP GET, PUT etc. which most web services do, against web arch anyway? So does this matter?


option 0: do nothing - engage in discussion with the TAG: 3 votes

<vikas> option 1

Option 1: the TAG text: 6 votes

<prasad> option 2

Option 2: the new text (Mark N's latest version): 5 votes

<scribe> Chair: who cannot live with option 0?

TomR: note that the TAG text concedes that going another way may be beneficial

<scribe> Chair: cannot live with option 0?

Total of cannot live with option 0: 6 people

Total of cannot live with option 1: 4 people

Total of cannot live with option 2: a few (with exact text)

<scribe> Chair: there are suggestions to modify TAG text a little.

Modifying TAG text in small ways

interactive wordsmithing

<mnot> Chair: Mark Nottingham

<scribe> Scribe: Tony

Jonathan: still have a problem with the phrasing of "deploying resources not on the Web"
... makes it sound like you lose the benefits of the other items that are on the web

Straw poll: who wants the phrase "not on the web"?

Total: 3

who thinks we should not use this phrase?

Total: 2

Who cannot live with "on the web"?

<scribe> Chair: we'll fall back to using "on the web" for now.

Latest iteration of the text:

<mnot> Web Architecture dictates that resources should be identified with URIs. Thus, use of the abstract properties of an EPR other than [destination] to identify resources is outside the scope of the Web Architecture. In certain circumstances, use of such additional properties may be convenient or beneficial. When building systems that use non-URI identifiers, care must be taken to weigh the tradeoffs inherent in deploying resources that are not on the

<mnot> Web.

We'll leave CR10 at that point until DaveO is present

Lunch - back about 13:10 local time

<vikas> I am here

<vikas> are you guys on mute

<Katy> Overload UsingAddressing marker to indicate async=full regardless of binding/protocol.

<Katy> New proposal for issue 59

<Katy> Overload UsingAddressing marker to indicate async=full regardless of binding/protocol.

<uyalcina> In the case of the WSDL SOAP/HTTP synchronous binding for request-response operations, the presence of the wsaw:UsingAddressing element does not change the requirement that the response message be sent over the same HTTP channel over which the request was received. In this case, the wsa:ReplyTo header in the request MUST NOT contain an address with a value different from the anonymous URI.

<dorchard> what's the uri for this proposal?

<uyalcina> Basically, we are proposing to change this paragraph to indicate that UsingAddressing semantics.

<dorchard> From the F2F agenda page at http://www.w3.org/2002/ws/addr/5/nov-f2f.html

<dorchard> Teleconference Bridge

<dorchard> TBD

<yinleng> MarkN: 3 proposals, one from Marc, one from Dave, and one from Umit

<uyalcina> umit+katy

<yinleng> Scribe: yinleng

<dorchard> What is the call-in #?

<GlenD> Zakim

<dorchard> #addr?

<dorchard> aha!

Marc: Presenting his proposal

Umit: Question on response binding.

Gil: Question on example v.

What is the significance on the Aysnc only element?

<dorchard> I lost everybody.

DaveO: E.g. v, two binding elements. Don't understand why there needs to be two binding elements to be used.
... Why do you need the second binding?

Marc: Bindings can specify a lot more than SOAP transports.

<TonyR> Glen: it would be nice to offer the ability to "derive" much of the settings from the original

<GlenD> Dave, the idea is that you might need to express MORE than just the SOAP transport binding... perfect example is what if I want the response to be delivered with the non-SOAP WSDL HTTP binding....

Anish: Made a suggestion that there are 3 different parts in this proposal.

DaveH: Seems like there are 2 proposals.

Hugo: What if your binding supports only 1-way messages? Want to send a message back. What do you do?

<TonyR> DaveO: question: looking at text here, doesn't seem to define sync / async

<TonyR> Glen: Anish was getting at that, too

Marc: Will use DaveO's definition of async.

<uyalcina> Anish has articulated my biggest problem with respect to the proposal.

<TonyR> Anish: did Marc consider correlating two one-way messages as an alternative to the strangeness that we see here?

Jonathan: Do we have consensus that we need to solve this problem of HTTP one way, and SMTP the other way?

<TonyR> Marc: didn't want to limit the solution to SOAP over HTTP - wanted to accommodate other protocols as well

<uyalcina> The charter does not require us to solve this

<dorchard> I think that multi-protocol is optional.

<TonyR> Marc: can do this with extensibility, but not keen on leaving it all to extension

<TonyR> Chair: everyone understands Marc's proposal?

<mnot> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Nov/0014.html

<TonyR> DaveO presents his proposal

DaveH: We need to just allow for one protocol in and another protocol out, but need not limit it to HTTP and SMTP.

<TonyR> DaveO: note that there is no mention of asynchrony, nor of particular protocol bindings, in the description (3.1.1)

<TonyR> DaveO: using MEPs to avoid specifying things at the protocol level - this makes it protocol independent

Anish: Always never defined in terms of MEPs. Are you discussing in terms of SOAP MEP or WSDL MEP?

DaveO: Intended to be protocol MEPs
... Abstraction of protocol MEPs so that they are protocol-independent.
... request-optional response MEP is invariant of whether SOAP 1.1. or 1.2
... Example 3:5 shows accommodation of using HTTP one way, and SMTP the other.
... For the ability to switch response protocol, this Responsebinding will do it.

Hugo: Can you accommodate SOAP request on input, then response binding to XML over HTTP, but not SOAP?

MarcH: Maintain the QName way is more flexible than URI
... The SOAP over SMTP binding is written in terms of SOAP MEP. Is that one way or is that request-response?

<dorchard> [1] http://lists.w3.org/Archives/Public/public-ws-addressing/2005Jul/att-0010/ws-addr-soapadjuncts-simplemeps_httpbinding.html

Take a 15 minute break now. Come back at 2:45pm.

Resuming now. 2:55 pm

<mnot> 4th proposal for i059:

<Katy> proposal 4: UsingAddressing marker indicates that supporting full WS-Addressing capabilities

Umit: Presenting proposal.

<mnot> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Oct/att-0116/ProposalTake3.htm

<vikas> +1 on this proposal

Anish: Using addressing, in the SOAP over HTTP case, is there any implication wrt to where the addressing marker occurs?

Hugo: You are not talking at all about response binding?

Umit: Yes, we are responding to one of the three parts.

Jonathan: Is this behavior of the marker tied to the SOAP binding?

<anish> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Nov/0026.html

Umit: With this proposal, this clarifies the SOAP over HTTP binding.

Anish: presenting his friendly amendment.
... This proposal is in the spirit of composability.
... Defines async as a hint to the wsdl processor that the request and response may be separated by large time interval, whatever large time interval means.
... Hints are important to know whether to generate a blocking model. Hints may be ignored.
... Proposal adds to Marc's propoal. Also adds to Umit's and Katy's, but not DaveO's which already covers it.

MarkN: This issue is more complex than "which proposal do people prefer". Should look at functionality provided by each proposal.

Discussion now on Requirements.

Resumption of questions for DaveO on his proposal as DaveO has rejoined.

<dorchard> the phone just muted.

Anish: How does one capture, in the SMTP binding, the property on the binding?

DaveO: This propposal is not meant to cover that scenario.

DaveH: How would you characterize the differences between your proposal and the other proposals?

Glen: Why do you need shim MEPs when you have got WSDL MEPs? Aren't these the same?

DaveO: No. Hoping to get rid of SOAP MEPs.
... Problem is SOAP MEPs as deployed in SOAP 1.2 are very constraining.

<dorchard> http://lists.w3.org/Archives/Public/public-ws-addressing/2005Jul/att-0010/ws-addr-soapadjuncts-simplemeps_httpbinding.html

<TonyR> DaveO describes how he is eliminating the problems of SOAP MEPs through his new MEPs

<TonyR> DaveH: so you have both WSDL MEPs and protocol MEPs, both of which can be non-trivial

<TonyR> DaveO: yes

DaveH: Is this a discussion we should have, how many levels of MEP, etc. ?

<TonyR> DaveO: big advantage is eliminating the complexity of the SOAP MEP state machine

<dorchard> daveh, a key difference between the proposals is how many levels of meps, so how many levels is by definition in the discussion.

Resuming discussion on "i059 Decision Points/ Requirements"

1. Does async need to be controllable at the abstract level?

<uyalcina> Dave I just sent you the proposal 4, you are ahead of the group.

2. Does async need to be controllable at the binding level>

3. do we need the ability to specify binding details for responses?

4. What granularities (message, operation, etc.) does async need to be controllable at?

<dorchard> got the proposal, thanks umit!

5. How do we express async in terms of SOAP / Underlying protocols?

Change 4 to 4. What granularity of attachment (message, operation, etc.) does async need to be controllable at?

6. What async modes do we want to enable (always, never, optional, etc.)?

Removed 2.

<dorchard> 2: binding uri; 3: binding qname. :-)

Replacing 2 with 2: Do we need to be able to enumerate possible underlying protocols for responses?

Replace 3 with 3. Do we need the ability to parameterise binding details for responses?

7. Should the semantics of usingAddressing be defined in terms of a specific WSDL Binding (e.g. the SOAP binding)?

<dorchard> DavidHull, I think there is a question about what is the dependency between the usingAddressing and the contained binding or a referenced binding within usingAddressing.

8. Do we have to define complete async semantics for anything beyond SOAP /HTTP?

<mnot> cr10 proposal:

<mnot> Web Architecture dictates that resources should be identified with URIs. Thus, use of the abstract properties of an EPR other than [destination] to identify resources is outside the scope of the Web Architecture. In certain circumstances, use of such additional properties may be convenient or beneficial. When building systems that use non-URI identifiers, care must be taken to weigh the tradeoffs inherent in deploying resources that are not on the

<mnot> Web.

<dorchard> Thus, use of the abstract properties of an EPR other than [destination] to identify resources results in resources that are not on the web.

<dorchard> Thus, use of the abstract properties of an EPR other than [destination] to identify resources results in resources that are not on the web.

<dorchard> Thus, use of the abstract properties of an EPR other than [destination] to identify resources results in resources that may not be on the web

<mnot> Web Architecture dictates that resources should be identified with URIs. Thus, use of the abstract properties of an EPR other than [destination] for identification may result in resources that are not on the Web. In certain circumstances, use of such additional properties may be convenient or beneficial. When building systems that use non-URI identifiers, care must be taken to weigh the tradeoffs inherent in deploying resources that are not on the

<mnot> Web

RESOLUTION: Approved minutes of the Oct 31 meeting.

Discussing the answers to the questions posed in the "i059 Decision Points/Requirements"

<dorchard> DaveO's proposal: 2: yes; 3: no; 4:? 5: at the protocol mep level; 6: all modes; 7: no; 8: yes.

Discussing answer to question 2

Straw poll on question 2

<vikas> abstain

Yes - 7

No - 3

Q2 result - YES

Discussing answer to question 3

Straw poll on question 3

<vikas> yes

Yes - 5

No - 3

Yes - 6

<dorchard> who's voted no?

Q3 result - YES

Discussing answer to question 4

Replacing 4 with 4. Does async need to be attachable at operation level?

Straw poll on

Yes - 7

Rest are abstention

Discussing answer to question 6

Q4 result - YES

Replacing 6 with 6. Do we want the service to assert control over synchronicity?

<dorchard> there's 6a) service can say anonymous only; 6b) service can say non-anon only; 6c) complete client control (status quo).

Straw poll on 6

Yes - 5

Yes - 7

No - 5

Q6 result - yes (?)

Discussing answer to question 7

Discussin answer to question 8

<dorchard> I believe we must talk about words like "The “full” value indicates that either one or two protocol message exchange pattern (proposed in [1] request-optional-response MEP or in [2] as SOAP-request MEP) instances is used to transmit the request and response. " now.

Straw poll on 8

Yes - 2

No - 2

Has this discussion enlightened anyone or confused?

Straw poll on question 1

Yes - 3

No - 3

mnot: Try to eliminate at least one proposal tomorrow.

<vikas> are we done?

<vikas> bye

Meeting adjourned till shortly after 9 am tomorrow.

<dorchard> ogenki des ka?

<dorchard> Hai, so des ne?

<Katy> scribe: Katy


DavidH: Suggested slight change to 2nd sentence 'may result in the resource not being on web'

Hugo: TAG were stressing that Web architecture should use URIs to identify. New version is too watered down and lost TAG message
... Would like to see Web architecture mentioned in 2nd sentence - to indicate use of EPR abstract properties to identify resources outside scope of web architecture

(this is what was in previous version yesterday)

Jonathan: agreed

David: Against Hugo's proposal - where is the 'scope of the web architecture' defined

dorchard: this is not outside the scope of the web architecture - the term 'on the web' is preferable as it is a defined term which we can concretely say

<dorchard> http://www.w3.org/TR/2004/REC-webarch-20041215/#uri-benefits

jonathan: If Web architecture is a spec then it's resaonable to define Web architecture but otherwise more abstract definintion

dorchard: 'on the web' is more specific - not dealing with abstract notions. Can live with stronger statements though
... 'on the web' = URI which is gettable

jonathan: Concern of TAG is non-HTTP gettable resources

Jeff: +1 for initial proposal - clear

Paul: TAG are doing a lot of work for us which will bring benefits. Would like to change wording slightly to focus on identification with URI vs identification with EPR

Mark: HTTP forms has same issue with identification but isn't 'contrary to the web architecture': Is this the issue here?

<scribe> Chair: Mark

General discussion on how to say EPR abstract properties lose benefits of Web architecture

<abbie> can u have the proposals on irc

4 proposals:

1 . From TAG

2. ... Thus, use of abstract properties .....for identification may result in the resource not being on the web...

3. ... identify resources is out of scoper of the Web Arch

<dorchard> What the heck, I'll vote for #1

4. ... id resources loses some benefits of the web arch

4. ... id resources loses core benefits of the web arch

<prasad> There are two (4.) above. Is that last one a 5.? or the real (4.)?

<TonyR> last one is the real 4

<vikas> vote for Proposal 2

<prasad> I vote for 4

Straw poll result: 1)=4 2)=4 3)=2 4)=7

<scribe> Chair: Can anyone not live with the 4th one?

<scribe> Chair: Take away and consider: we will chad later


<scribe> Chair: Summarised position from yesterday

<Gil> Please may I make a suggestion which might help us reach swifter resolution to the issues tomorrow?

<Gil> We have 4 possible proposals on the table:

<Gil> 1. (Initial SAP/Oracle/IBM proposal) Superceded

<Gil> 2. Marc's proposal based on 1+ Anish's friendly amendment

<Gil> 3. David's proposal

<Gil> 4. Umit's/my proposal

<Gil> Rather than discuss the above as complete proposals, could we separate into composable functions and discuss as such? I.e.:

<Gil> A. Specification of async only or sync only at binding level

<Gil> Should we opt for:

<Gil> asyncOnly attribute from proposal 2 OR wsaw:Async from proposal 3 OR no additional flag from proposal 4 (OR other)

<Gil> B. Specification of binding for async response

<Gil> Should we opt for

<Gil> wsaw:Async binding semantics from proposal 2 OR wsaw:ResponseBinding semantics from proposal 3 OR no response binding specification (OR other)

<Gil> C. Specification of async at the interface/op level

<Gil> Should we opt for

<Gil> Anish's friendly amendment to 2 OR no specification (OR other)

<Gil> There may be finer points that I have missed, but this structure might help focus the debate.

<Gil> Many thanks

<Gil> Katy

Umit: Compose complete proposal based on function

<bob> ping

<abbie> no

dorchard: Concern that it's not possible to separate A B and C above

<charlton> WG reviewing 4 proposals on the table vis-a-vis the i059 questions

Glen: A B and C will give framework

dorchard: agree but concerned that we may have problems with this approach

<Chair: A. Specification of async only or sync only at binding level

<abbie> +1 dorchard

marc: can't choose between asyncOnly and async without considering B

Katy: could we consider whether we need asyncOnly or synconly first?

dhull: change markup so we are saying 'I can acccept anonymous only'
... clarify concept - anonymousOnly (sync) NoAnonymous (async)

Tony: capability of the server rather than anonymous - tying to anon not good

Katy: Like to see requirement for sync only or async only

Anish: don't agree with Tony

<abbie> +1 Anish

Glen: leave async as that's what we understand.
... use case: sync - provides mechasism for partial support without having to have multi thread

use case: async only enables the service to never block

<Gil> +1 glen

Tom: sync only: migration use cases

dhull: Agrees with Tom and Glen

anish: don't think this is critical functionality. Issue is: is the service required to specify

dorchard: Service should be able to specify level of service that it supports. Currently we are client centered and this provides a mechanism for service to define its behaviour
... This is a useful mechansism to control behaviour of service.

katy: proposal 4 is not saying that implementations must be sync only or async only - could still use a fault

glen: agreed but would still vot e for sync/async service definintionto save runtime round trip to establish service capabilities

jonathan: +1 to Katy (proposal 4)

<abbie> +1 dorchard

Anish: if we decide to go with marker - shouldn't restrict the marker to request response MEPs - being able to accept anonURI (or not) is useful for other MEPs

Umit: Fault proposal should have been in proposal 4. Not required behaviour for service at this point

Gil: Disagree with 4 because problem should be addressed during development time rather than at runtime

dorchard: +1 Gil whole point is to specify contract as full as possible

tony: +1 to above

Paul: undecided - doesn't need to be done in wsdl

<abbie> +1 dorchard

dhull: wsdl marker mechanism to prevent fault from happening

<scribe> Chair: vote no flag v flag

<dorchard> Can we record the votes into IRC?

Result: YES (want flag) - 9 NO (no flag) - 6 Abstain - 2

<charlton> No formal objections

RESOLUTION: A Specification of async only or sync only on WSDL is REQUIRED

B - Specification of binding for async response

dhull 4 Options:

dhull 1)do nothing;

dhull 2)protocol/transport;

dhull 3)WSDL binding;

dhull 4)or combine 2 and 3

tony +1 to DavidH

Marc Need to be more consistant about terminology with wsdl. Make sure that we need transport (not protocol)

dorchard agreed with dhull

dorchard Note sent this morning...

dorchard: Speaking for 1 or 2. 3 makes common case hard

marc: 3 is more capable, 2 is ok: 3 provides more than just transport binding

glen: In general what we want is SOAP - most issues are wrt how to bind to SOAP (e.g. what are the fault codes going to be) not WSDL binding level.

glen: speaking in favour of 2 - don't require anything as complicated as 3

hugo: would like to understand if complications switching bindings

umit: 2 will be half-baked and cause more problems. Favour of either 1 or 3. Not 2

dhull: Clarify meant 4 to mean: 3 + some convenient mechanism for 2

anish: 2 does not make simple case simple - makes assumption that there is no variability in the wsdl bindings: Vote 1 or 3

paul: speak for 2

marc: 1 is an interoperability problem. People will still want to return async on different transport. Not specifying relationship in wsdl will cause problems.

dorchard: There are a lot of things that can be specified in a binding - but 3 brings in additional complexity if there are conflicts between request and respnse

dorchard: Number 2 solves majority of cases. Small step which will get us partially there.

glen: 2 could be used with additional property values for JMS (eg) specific values

glen: 2 is a good step at this stage

hugo: Concerned about impact on schedule. Could we put this in a separate document?

umit: +1 to hugo

chair: possible of 2.0 or separate doc for schedule

gil: In favour of 1. Accept that there will be proprietary solutions but it would be worse if we overshoot schedule.

jonathan: fear unexpected interactions with supporting 3

marc: please could david explain use of component designator syntax?

dorchard: could possibly put in wsdl 2.0 binidng uri but behaviour will be undefined under this proposal

dhull: clarify 1: is this same protocol vs async over http?

glen: this is a separate issue - we are just discussing whether we have wsdl markup for async response

paul: Focus on the schedule. WSDL binding will old up. Long time spent discussing this already - don't spend more time on this.

umit: in charter, nothing to deliver response binding. speaking in favour of 1.

david: if do 3 might as well do 4

dhull: leaning towards 1. don't commit more than required for working group.

dhull: willing to opt for 1 but don't want to box in wrt adopting it.

gil: 1 does not restrict asnc behaviour to underlying protocol

Chair: chad: please express choices in preference

Chair: option 1 - do nothing.

Chair call for objections: none

RESOLUTION Specification for async response NO ACTION

C Specification of async at the interface/op level

Anish This proposal is addressing the core issue of PM async/sync capabilities - not relating to binding

dhull this is a hint - no further semantics. Is attaching this hinr promising that we will support anonymous URIs?

anish the marker has no implications wrt value of MEPs or service support of anonymous. All it is saying is that there might be a long delay between request and response

jonathan separate from WS-Addressing - this is a WSDL issue.

anish to me, this is really the async issue. the issue is core to the client PM - developers aren't concerned with underpinning transport/binding issues

dhull Names are important again here. Callbacks at client level is important but async is more about freely flowing msgs not request response.

dhull This seems out of scope to talk about code generation in the context of ws-addressing

umit observation that this appears to be a wsld problem - it is more of style hint to help tools to generate programming model.

jonathan explained style hints in wsdl 2.0: you can use this hint. Doesn't this require more force than a hint? I.e. what is the result if I ignore the hint?

anish Would like this to be a hint - it's not possible to specify the details. To ignore the hint will be at implementor's peril

glen should this be a policy assertion not boolean.

glen this is not hard to do and is not critical to the WG at this point.

anish motivation for why this dosn't have implications on anon/nonanon. this is because this marker does not know what anonURi means to the underpinning binding.

dhull Anish's point is very helpful. This clarifies that, at this level there's nothing addressing specific that we can decide.

dhull I don't think this is in scope.

dorchard what is the relationship between this marker. WSDL request/response wrt underlying protocol

dorchard Also what's relationship between partner links and service links - i.e. 2 one way correlated messages

anish this marker is making no statement of protocol level just saying to the client that request response is separated by a large amout of time.

anish BPEL partner links question: because this is such a common problem, this is not sufficient soltion

jeff +1 to above

jonathan This is not a length of time between request and response - this is a client programming model hint.

umit agrees that programming model characterisation

gil If abstract interface level uses this markup but binding is async=never, should wsdl processor flag an error?

anish the async never is talking about the value of the wsa:replyTo so this is slightly different.

jonathan i think that this proposal is interesting but whether we do this in this group is an issue.

dorchard I heard that the flag has no affect on the other WS-Addressing elements of attributes in the binding. I am against any thing that is just a hint because we wont be able to test. With constraints on binding this would make this more applicable.

anish I would accept an amendment for restrictions on bindings. Also, it should be a positive point that there is no impact of this in testing.

umit response binding was far simpler to this because it's just a programming hint.

jeff we keep saying that we will do things elsewhere.

paul if this isn't a testable assertion, then I don't see the value in it.

dochard I think that there are 2 possibilities hint only or binding implications

chair chad - 3 options:

a) no

b) yes - as a hint as initially described by Anish's proposal

c) yes - abstract assertion with constraints on binding

winning vote a

RESOLUTION: Specification of async at the interface level - NO ACTION

Specification of WSDL MEPs to SOAP/transport-level MEP mapping

<mnot> 1.We don't model this at the SOAP level.

<mnot> 1.We don't say anything at all about it.  We talk about the WSDL binding, we mention the 202 rule in the context of SOAP/HTTP, and that's it.

<anish> Scribe: anish

<Jonathan_Marsh> We mention the issue, and mention that there are several possible approaches (see below), possibly include an editorial note asking for more input in LC, but don't commit to a particular approach.

Vote on whether the WG wants to go into specifying mapping of WSDL MEPs to SOAP/transport-level MEPs

<Jonathan_Marsh> 2. (DH1) We say that each physical exchange constitutes a complete SOAP MEP.

<Jonathan_Marsh> * If there is a fault, then the HTTP exchange is an ordinary SOAP request-response, with the fault as the response

<Jonathan_Marsh> * If there is no fault, then the HTTP exchange is still an ordinary SOAP request-response, with the 202 modeled as an optimized representation of a SOAP message with an [action] marking it as an acknowledgment. The email message is a SOAP one-way.

<Jonathan_Marsh> 3. (DH2) We say that each physical message is a one-way SOAP message exchange, distinguishing "SOAP/HTTP request" and "SOAP/HTTP response" as separate bindings of a SOAP one-way message exchange.

<Jonathan_Marsh> * If there is a fault, then there was a "SOAP/HTTP request" message and "SOAP/HTTP response" message.

<Jonathan_Marsh> * If there was no fault, then there was a "SOAP/HTTP request" message and a "SOAP/email" message. The 202 follows from a general rule of "send back 202 if you don't have anything else to send back."

<Jonathan_Marsh> 4. (UKA, if I understand it) We say that, in the context of async WSA, SOAP/HTTP can mean either what it already means, or it can mean send back a 202 and send the actual response some other way. We specifically define what this means if the response is sent through a separate SOAP/HTTP interaction, but not for other transports (i.e., we don't explicitly describe an email response).

<Jonathan_Marsh> 5. (DO, as I understand it). We model the HTTP exchange as an instance of a transport-level in-optional-out MEP, and the email (if it happens) as a transport-level one-way.

<vikas> vikas: Don't go there

Vote, don't go there: 9

go there: 2

abstentions: 4

Mark: we won't go there, lets look at #1

DaveO: i'm surprised that we are going to define a new binding

Glen: yes or modify the existing binding

umit: if you want to pursue option 2-6 you can still do it

glen: the problem is that when i want to do this with email then i have to do something else

anish: why is this a problem for SMTP

glen: cause i want to be able to compose one-way SMTP binding for req-res

marc: option 1 seems a bit odd, it says we don't say anything about it and then we go on to say what we are going to say
... we say things about 202 etc

daveh: by we don't say anything about 'it' I mean that we don't say what happens in all cases but only in the SOAP/HTTP case

marc: don't know how to include this in the wsdl-binding doc?

daveh: at sometime someone should spin up a framework. we are not prohibiting someone from defining a problem
... its just that we haven't nailed it down and therefore shouldn't be in the REC

Mark: do we have enuf information to close i059
... if so, i want to give an AI to someone to come up with a proposal

umit: proposal 1 (initial proposal) already addresses our decisions that we made today

anish: except the proposal is a soap1.1/wsdl1.1 specific thing

Mark: lets not get into syntax

hugo: we have another option to not say anything at all

Mark: the options are 1) we talk about 202 etc stuff, 2) 1 + add a ed note, 3) do nothing at all

daveO: i don't understand how the WG could realisticaly thing that it should define SOAP binding extension and smuggle them inside the WSDL extension. It should be a separate doc for this.

<uyalcina> It will take a while for Anish to scribe what he has just said, but a big +1

anish: would like to speak for 1. Keep it simple.

glen: mark the extension as required and that would be OK to do that.
... there is a spectrum between defining a 202 and defining a whole new binding
... would like to speak for 3

<uyalcina> i am very against option 3, we would have to define this in the testing framework. We will basically bless the behaviour in the spec if we go for 1. This is essential for interop,

gil: i'm a developer and dealing with WSDL and you want my tools to associate WSDL with my SOAP stack?

Marc: yes

glen: you already do that

anish: we define the extension and spell out exactly how it works for soap11/wsdl11 and let XMLP figure out their new binding for SOAP

Mark: are we ready to make a decision of 1 v. 3
... anyone object to 1?

One objection

Mark: lets take a vote

<vikas> vikas: Vote for 1

option 1: 5

option 3: 2

abstain: 8

Umit: this particular behavior is already in the test suite

Mark: it was asserted that all the decisions that we made are in proposal 1, is that true umit?

umit: yes

Mark: lets figure out the details on concall later and discuss testing for the rest of the f2f






Mark: we left off with original proposal 1
... umit can you walk us thru

umit goes thru the proposal

<discussion on umit's proposal>

hugo: in the proposal, you alternate between soap/http and soap1.1-http. which part applies to SOAP 1.1 and which part applies to SOAP 1.2?

umit: only applies to SOAP 11/http only. There is an open issues section

hugo: my 2nd question is: in the 'always' case, there is a MUST. Don't that mean that you need a wsdl:required='true'

umit: yes, probably

Mark: to move this forward, we need to refresh the proposal. Can umit/katy take an AI for that?

umit: is the request to rewrite it to make it more general or is it a rewrite for the next step?

hugo: i thought the attribute applied it to more generic and specify details for soap

Marc: if not we need to make the name more specific to http/soap rather then general 'async'

Mark: another thing is to start talking about SOAP 1.2 and WSDL 2.0

Umit: if you have specific problems with the text then pl. point them out

Mark: get a rewrite and then start generating issues about it, eg: do we need another fault? how to deal with WSDL 2.0/soap 1.2
... once Umit gets the new doc out, pl. identify the issue in the subject line of the email

<scribe> ACTION: Umit to rewrite proposal 1 [recorded in http://www.w3.org/2005/11/07-ws-addr-minutes.html#action01]

<scribe> ACTION: Umit to rewrite proposal 1 by 2005-11-16 [recorded in http://www.w3.org/2005/11/07-ws-addr-minutes.html#action02]


Mark: is there anything we can do to move along SOAP 1.2/wsdl 2.0

hugo: we don't need to do anything about wsdl 2.0 and for soap 1.2 has a dependencies on XMLP

Mark: i would like to close this down soon, so review Umit's rewrite
... we only have i059 left. we have made significant progress on it. We have Core/SOAP in CR.
... we have closed all the known CR issues
... we may get one from the TAG

Umit: the SOAP faults should really live in the SOAP binding doc

Anish: there is a note in the SOAP binding CR doc about additional faults may be defined/changed. I hope that will allow us to define a new fault without going back to LC

hugo: i probably be ok

Mark: umit, why don't u file this as a separate issue
... to get out of CR, we really need to get this testing done
... my inclination is to still have an interop event
... when we are done with WSDL stuff we may have another interop
... some people may balk at 2 different interop events
... people can choose which interop even to attend
... vancouver interop/f2f is the week of 16th jan: tue-friday noon
... I hope to go to PR after the interop
... if we can be done with i059 we can get an LC doc before the holidays

<scribe> ACTION: Umit to file a CR issue about SOAP faults [recorded in http://www.w3.org/2005/11/07-ws-addr-minutes.html#action03]

<vikas> i am here


Paul: the testing effort is represented by the testing doc itself, which is quite up to date
... i think we should address is the list of testable features
... we don't have a feature about the metadata bucket
... we have test for all the the feature except core07, core08 and soap07

<pauld> walks through testable features: http://www.w3.org/2002/ws/addr/testsuite/features/

<pauld> http://lists.w3.org/Archives/Public/public-ws-addressing-tests/2005Oct/att-0007/WS-Addressing_Test_Cases.htm

Paul: the feature list at http://www.w3.org/2002/ws/addr/testsuite/features/ is for soap/core doc right now
... there is no reason not to add additional tests if we can all agree to it

Mark: we can add more test and say that some test are required to exit CR and there could be more tests which raises the bar

Paul: there is an issue about IRI comparison tests

<vikas> going to drop off

Daveh: is this a testable feature

mark: can we drop core08 as a testable feature?
... we need a flag about whether it is core/soap or wsdl
... can we have a XSL view for various facets?
... lets drop core08 as a testable feature, we can still have the tests

Paul: we need something for optional faults

Mark: we need to have tests for fault semantics
... we need 2 impl. of this to get out of CR

daveh: do we have a test for each of the fault?

Mark: no

Paul: each test requires an example XML message and an xpath expression associated with it

Daveh: i had envisioned that u send a req that has a reply endpoint and that reply endpoint has some sort of complex context for which u should get a fault back

<jeffm> scribe: jeffm

mark: does anyone want to do the work on faults
... who will implement these faults?
... if two do not, (they are optional features) then we will have to consider dropping the feature from the spec
... wants to know if anyone knows at this point if they will implment these -- no answer yet

paul: each test has a "cost" -- need to be written and then reviewed to make sure the test is correct
... need to have at least one test per fault

mark: if it proves to be too difficult to test particular features, a possibility is to move that feature from core to a note

umit: volunteers to write some tests if paul will help to review and instruct
... why don't we split up the tests across wg members

daveh: doesn't quite buy have to have one test for all -- e.g. endpoint not avail --- is that really testable

paul: hugo submitted a test case -- one way message with none value for fault, but get a fault due to the way soap processing model works
... this part of value of cr, get feedback on fault design

mark: need to wait and see if any tests surface
... question: core07 -- will that be hard to develop a test?

paul: no -- just overlooked it

mark: strawman to move forward - test subgroup reg call 11 am PST
... as test group believes that they have a test completed can they bring it up to the WG for approval

jmarsh - in xmlp only issues that the test group couldn't resolve or had questions on went to the WG for review

mark: wants to make sure we make incremental process on the test suites, not wait until "the end" to have pushback from the WG
... are people comfortable with leaving it to the test group, and have the WG mainly control the Testable Features list

jmarsh: keep that list to what is required for CR, other desireable testing features should go into a different bucket

mark: 4 buckets for tests -- CR vs. extra and core/soap vs. wsdl
... what is process for deciding when something gets on the "official list"
... heavyweght -- WG approves every, lightweight -- WG only approves if their is a question on test group
... we will go with the lighterweight so -- WG members should track what the test group does, and bring up questions first with the TG

paul: not sure when we will be ready to really test --

mark: would be nice to have a web page providing instructions on how to run the various tests
... of the cos who have indicated they are going to test, can they provide a date for when they might be ready to start -- by jan

indication that 3 or 4 cos, at least, will be ready

paul: has developed wsdl 1.1 and wsdl 2.0 defs of test service -- feels pretty confident of the 1.1 wsdl description

not as confident that the wsdl 2.0 version is quite correct

paul: don't know how to describe soap 1.2 in wsdl 1.1 doc

mark: test telecon will happen next thurs
... no ws-addr telecon next week -- mark will be on vacation
... there will be a ws-addr call the monday of the week of US thanksgiving
... looking forward to a concrete proposal for i59
... AOB?

march: asks about resolution to issue 65 -- he sent an email
... 3 nov
... at end of email: subj Stable WSDL Binding Draft Available
... issue 63 (not 65)
... pushes back on issue -- will complicate the doc and require a lot of duplication

glen: why can't u define the xml mu -- and say in wsdl 1.1 do it this way, in wsdl 2.0 do it this other way

hugo: this methodology should be acceptable -- doesn't see any problem

no objection

mark: looking for a "bunny"

<scribe> scribe: you ask him what that means :-)

mark: bunny's task is to summarize the meeting results for wsdl WG f2f

glen: asks why the chair wouldn't do that?

mark: not my job
... is wsdl doc uptodate except for today's decision and i63
... asks katy and umit to present their propsal --- the price of success ;-)

mark twists their arms --

they agree

the crowd cheers

march followed by katy/umit will summarize the meeting for the wsdl jt f2f tomorrow

thanks to hitachi for hosting


especial thanks to nishiyama-san

<more loud applause>

Summary of Action Items

[NEW] ACTION: Umit to file a CR issue about SOAP faults [recorded in http://www.w3.org/2005/11/07-ws-addr-minutes.html#action03]
[NEW] ACTION: Umit to rewrite proposal 1 by 2005-11-16 [recorded in http://www.w3.org/2005/11/07-ws-addr-minutes.html#action02]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2005/11/27 06:39:43 $