W3C

Web Services Addressing Working Group

22 Aug 2005

Attendees

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

Contents


<scribe> Scribe: Prasad

Date: 22 Aug 2005

Agenda review, AOB

Mnot reviews the Agenda.

Additional agenda items:

1. the issue Jonathan raised today, "Fault: interaction with delimiter".

2. Mnot adds, discussion of WSDL document drafts after agenda item 5

(future meetings).

Call for corrections to the minutes

- 2005-07-25: <http://www.w3.org/2002/ws/addr/5/07/25-ws-addr-minutes.html>

No corrections.

July 25 minutes approved

Review Action Items

2005-07-18: Jonathan Marsh to coordinate with Description to make

sure notational conventions are in sync.

Mnot: Jonthan raised with WSDL WG but they did not adopt it as it.

Jonathan: WSDL has different clases of extensions. So more discussion is

needed. So notational convetions are not aligned but

it is recorded a last call issue with WSDL. We can close this AI as I

raised the issue.

AI recorded as CLOSED.

2005-07-25: Paul Downey to generate list of features, their

requirement level and applicability for discussion.

<scribe> PENDING.

Future Meetings

Mnot: We have two F2F meeting End of Sept in Bay Area hosted by TIBCO.

Registration open.

Early Nov in Tokyo hosted by Hitach. We don't have arrangements page yet

but will have soon.

Registration open for both.

Longterm Likely to have a f2f In Jan 06. Looking for a host in US. If

interested get in touch with MarkN.

Another in late Feb or early March '06 during Tech Plenary.

Telcon this and next week but off for one week due to labor day holiday.

WSDL document Drafts (Additional Agenda item)

Mnot: We are required by the W3C process to publish a heart-beat

document every 3 months.

WSDL WD we have not published an update in a while, been more than 3

months already.

Do we need to publish the draft anyway saying we don't have any changes

since the last draft?

Hugo: We have never done that before. We need to publish something every

3 months.

Mnot: My understanding is that a Working Draft once published it needs

to be heart-beated.

Jonathan: When do we expect to get through the remaining issues on the

WSDL document?

Mnot: I guess my question is can we publish an update soon enough?

Anish: Looking at the open issues, other than issue 59, all have pending

proposals. We should be able to resolve quickly.

Glen: Even for 59 there are proposal but no consensus yet.

Jonthan: I have another issue I am contemplating to file but it is

mainly editorial. I think we are pretty close.

We also have editorial copies publicly available. Republishing same

again seems silly.

TomR: regarding 59, could some explain if one of the possibilities not

to do anything?

Mnot: We should discuss that when we discuss issue 59.

Hugo: This rule is to ensure that the WG is doing some work in the

publicly visible area.

I don't see a problem in not re-publishing the same WSDL document again

esp. when we have made any changes.

Mnot: Ok. It seems we don't need to do a heart-beat at this point.

Issue i041 - Use Cases for Testing

Mnot: Opened a long time ago when we were still in WD.

This is not a WD issue anymore. Can we close it or move

it to CR issues list. Is that ok with people.

Glen: As long as we don't drop anything on the floor.

Mnot: We know we need to do it anyway for CR.

Mont: Any objections to closing?

None.

Recorded that issue i041 is Dropped.

Mont: Paul does not have connectivity so, it might be couple of weeks

before we will be able to start

discussing it. Perhaps we should assign the action to someone else?

Mnot: Can someone take on the action to come up with a list of features

in the spec

that are in CR. What kind of requirements they entail. The

applicability of how they might be tested in different scenarios.

Tested on client or server side etc.

Jonathan: Paul was going to come up with a test assertion doc. I am

interested in updating the scenarios we submitted and updating them to the

CR draft. I think that is different.

Mnot: We just need a list of testable features and different ways they

can be tested.

Jonthan: In Xquery group they have separate co-char for driving the

testing activity of the WG.

Mnot: We don't have an issue with time. We have too little discussion.

Are people interested in a task force?

TomR: Our colleagues might be interested. We may want to ask for

contributions one more round.

Mnot: We made the contributions request a while back

Jonathan: We we have a task-force would people be interested?

Arun: I can prepare / contribute a list

MNot: Can you do by next week?

Arun: Yes.

ACTION to Arun to produce a list of testable features.

TomR: Is this different from CR testing?

Mnot: This is CR testing

Issue i017 - Purpose of the Action property

Mnot: This issue has subparts we closed parts relevant to Core but there

is one part that is applicable to WSDL binding

Anish: There were 2 subparts. 1st related to Core. 2nd related to WSDL.

The 2nd part is about operation name mapping of WSDL 2.0.

It was ONM requirement. It is no longer a requirement but made a best

practice and move to WSDL 2.0 Primer.

If action property is that disambiguates messages we need way to state

that in WSDL.

There are five proposals. 1st by me is no longer relevant.

Jonathan: Since the last proposal was sent, things in WSDL changed. We

moved this from core as best practice to Primer.

The wording has gotten softer. It talks about how difficult to make all

your messages or even GED's unique.

Why isn't the information that all actions in the WSDL are unique via

inspection of the document? If you know

WS-A you can tell what the actions are either by defaulting or by

explicit description in the WSDL. Why do we need

to assert that all the Actions are unique?

Glen: This is similar to RPC-Style hint. I can look into to schema of

the operation and WSDL and figure out that it matches RPC-Style but,

having the hint there gives me head start. Doing the assertion at the

WSDL for Action gives me the flexibility to use them as the disambiguators.

Anish: We are getting into proposal discussion now.

1st option is do-nothing. You can determine the action is unique by

looking at the WSDL

The best practice says if we have multiple endpoints you should indicate

that via WSDL feature extension. But

we don't do that.

2nd option is the proposal I sent before we resolved issue 21.

It is now obsolete based on the way we resolved issue 21 and what is

proposed in the options 3 and 4.

option 3 is defining an @ called wasw:distictAction and allowing it to

be used on wsaw:usingAddressing.

scribe:

Mnot: Option 3? You mean Attribute or Element?

Anish: I switched 2 and 3

Mnot: So you prefer @?

Anish: Yes. Both do the same but I prefer @ as an aesthetic issue

Those are the options + Jonatha's option, where it is suggested to add

some text to WSDL and WS-A

Jonthan: explains his proposal what is now relevant given the changes to

WSDL.

Mnot: What do people think?

Paco: Agree with Jonthan. Using Action to disambiguate messages is

mostly implementation issue.

We should close with no action.

Anish: This is an optional marker. You are not forced to use it.

Mnot: Per the CR criteria we need to interoperable impl. for optional

features. We need to ensure we will have two such implementations.

Glen: Part of it is negative testing.

Mnot: Who is interested in implementing this.

Anish: Oracle.

Glen: I am

Mnot: Ok we have two.

Jonathan: Anish's preferred attribute approach does not give an option to

say this is required. wsdl:required or feature:required

Anish: When you use that attribute, you are asserting something

independent of WS-Addr. If you don't understand usingAddr

then it does not matter. If you do understand the usingAddr marker you

understand this marker also. If the value of this @ is true

then actions in WSDL have to unique or the WSDL is incorrect.

Jonathan: So if the marker is set to true, my WSDL / WS-Addr processor

must verify that all actions are unique and fault if not?

Anish: If that WSDL processor faults or not, is not an issue. We

discussed it earlier.

Jonathan: So, if the marker is there but all Actions are not distinct it

is not clear if the processor should fault?

Or is it still a discussion point. Trying to see if this is really

optional or not.

Aish: It is a best practice. We need to have discussion what sort of

failure we want to see.

Jonathan: We need to take that discussion to the list.
... There should be required marker on it, so that your processor

can reject.

Glen: As a client no issue. As a server........ This is really to help

dispatch

Jonathan: Do you agree Anish that this is really redundant info?

Glen: It is true, you can calculate it heuristically

Anish: we don't reject WSDLs without this marker automatically
... It is useful in the -ve case, when the WSDL designer intended

them to be unique.

Jonathan: We need to worry about markers in imported WSDLs etc.

Straw-poll who prefers the marker approach?:

Yes: 4

No: 7

Who cannot live with the marker?:

Jonathan: In certain scenarios

Who cannot live w/o the marker?: None

Mnot: giving the straw-poll, my inclination is not do the marker

We have two options: (1) we close the issue with no action

(2) Add some text

Anish: I prefer we clarify something in the spec

Mnot: Issue 17 to close with close with proposal two with Editors to

modify the text proposed by Jonathan

as appropriate.

No objections:

Issue closed as above.

Issue i020 Addressing and WSDL

Anish: describes the issue

Mnot: Perhaps we should let people to familiarize the issue and revisit

next week

Agreed.

issue discussion deferred to next week:

Issue i056 - Determining the value of the [destination] property from WSDL

Anish: Issue how do you determine the destination property from WSDL?

We were heading towards one of the proposals #4 or #5

Jonthan: What is the difference between these and #6 from MarkH?

Jonathan: Don't need to reconcile an EPR in WSDL and the URI pointing to

the same thing?

Anish: Doesn't that put us in EPR equivalence?

Jonthan: No.

MarcH: This is simply addressing how you derive the address of

destination from the WSDL to stick in the EPR

Aish: What happens when the EPR points to the WSDL......

Mnot: It seems we are leaning towards proposal 6.
... Anyone prefer anything other than proposal 6?

None.

Mnot: we will take it up next week.

i057 - Determining the value of the [reference parameters] property

from WSDL

Mnot: Does it make sense to resolve 56 before 57?

Anish: It makes sense

Jonathan: Reference params allow you insert headers. Other things in

WSDL allow you to insert headers.

Mnot: Reference parameters are not necessarily headers give our charter

needs us to be SOAP independent.

Jonathan: How do you reconcile same thing defined in WSDL headers and

Reference params in EPRs?

MarcH: I don't see value for refParameters in EPRs in WSDLs.

Mnot: We will take it up after issue 56 is resolved

i059 - Support for asynchronous / multi-MEP usage of web services

Mnot: This was handed to a task-force. They were almost done. What is

the current status.

Glen: I need to drop off now. Can we do it next week?

Mnot: This is the only real issue open against WSDL
... We will take up this issue also next week.

New issue from Jonathan this morning, "Fault:" interacts with [delimiter]

Jonathan: Describes the issue. Interaction of ":" delimiter with ":" in

Fault: "

Mnot: Did we visit this before?

MarcH: Prefer option 1, replace "Fault:" with "[delimiter]Fault[delimiter]"

Mnot: I will put it on issues list. We can discuss next week.

MarcH: Arun raised an issue on UsingAddressing vs Action that had not

been recorded.

Arun: Sent Aug 14th, with 5 proposals.

Mnot: Should we look at it now?

Jonathan: Like to review

Mnot: Ok. I will open the issue and put it on the agenda for next week.

Meeting Adjourned.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2005/08/23 21:21:29 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.127  of Date: 2005/08/16 15:12:03  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Plain_Text_Format (score 0.95)

Found Scribe: Prasad
Inferring ScribeNick: scribe
WARNING: No scribe lines found matching previous ScribeNick pattern: <Prasad> ...

WARNING: No "Present: ... " found!
Possibly Present: Aish Anish Arun Fault Glen Hugo Jonathan Jonthan MNot MarcH Mont Paco TomR
You can indicate people for the Present list like this:
	<dbooth> Present: dbooth jonathan mary
	<dbooth> Present+ amy

Found Date: 22 Aug 2005
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.


WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)


[End of scribe.perl diagnostic output]