Minutes WS Choreography WG Face to Face meeting 13-14th June 2005

Face to Face Meeting 13 and 14th June 2005-06-20
Sun Microsystems, London, UK

Roll Call
======

Charlton Barreto (Adobe), Paul Downey (Monday 13th only - BT - guest), Tony Fletcher (Choreology), Peter Furniss (Choreology), John Storey (Nortel), Martin Chapman (Co-Chair - Oracle), Jeff Misckinsky (Oracle), Nickolas Kavantzas (Oracle), Steve Ross-Talbot (Co-Chair - Pi4Tech), Gary Brown (Pi4Tech), Monica Martin (Sun), Yves Lafon (W3C staff), Kohei Honda (QMWC), Marco Carbone (QMWC - guest), Nobuko Yoshida (Imperial)


Admin
=====

IRC log at http://www.w3.org/2005/06/13-ws-chor-irc (only available for the afternoon session) and http://www.w3.org/2005/06/14-ws-chor-irc 

Scribes:  Tony morning of 13th, Jeff afternoon of 13th, Charlton morning of 14th and Steve & Charlton afternoon 14th.


Agenda changes:

http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/att-0004/Jun_05_F2F_Agenda_-_0.txt 
It was agreed to pick up on a fault handling after lunch in the hope that Nick would arrive by then first review outstanding issues - refer to Martin's email of Friday.  Add to the agenda consideration of the exit criteria.

Minutes for the last teleconference are not yet available but Steve said that no major issues were resolved.  One could look at the IRC log for the meeting (http://www.w3.org/2005/06/7-ws-chor-irc) and also Monica's draft minutes. 

Actions
======

1. ACTION: Martin to do UML diagram from scratch for CDL 
        NO PROGRESS - NO CHANGE
        Clarified that this is for the primer not the spec.

2. ACTION: chairs to respond to issue raiser for closed issues 
        STANDING ITEM 
Martin may need help sending out e-mails.  Still awaiting replies from outside the group, though he acknowledged he had had replies from group members.

3. ACTION: Steve and/or Martin to generate a list of the issue numbers that the editing team should be working on as a spot check.
        NO PROGRESS 

4. ACTION: Chairs discuss and provide a recommendation. 
        DONE - Mark as done as it is thought that this action was about using Bugzilla to track issue resolution into the document.

5. ACTION: SRT To add 1008 to agenda for F2F
        DONE

6. ACTION: Gary to modify proposal for 1001 changing session for instance
        DONE.

7. ACTION: Editors will take modified proposal for 1001 and wrap up into 2.4.7 based on amendment
        DONE - and now edited into the specification

8. ACTION: Gary to craft text for 1003
        DONE.

9. ACTION: SRT to add 1003 to agenda for F2F
        DONE.

10. ACTION: SRT to include 1039 in F2F Agenda
DONE

11 ACTION to review issues at face-to-face
        DONE -  next item on agenda.

11. ACTION: create an issue on activity ordering. 
DONE

12. ACTION: Gary to make a proposal on activity ordering
DONE

Outstanding issues
===============

A stock check of all open issues on bugzilla:
http://lists.w3.org/Archives/Member/member-ws-chor/2005Jun/0001.html

It was noted that issues 1001 and 1003 are on the agenda.

Issue 1101 The operational behaviour involving "complete condition" is not clear
There is a word change proposal - on agenda for tomorrow. 

Issue 1459 We should clearly state in the spec that nested isolation choreographies are not permitted
This issue has a proposed text change so yes we should discuss.

Issue 1004 Workunit Guard Condition Semantics
Gary and Steve confirmed that they were happy for this issue to be closed without further change as the semantics had been clarified at a previous face-to-face meeting.

Agreed to Close WONTFIX

NEW ACTION: Mark 1004 as Close WONTFIX

Issue 1005 Workunit Repetition
As for issue 1004.

Agreed to Close WONTFIX

NEW ACTION: Mark 1005 as Close WONTFIX

Issue 1021 Point of clarity - Section 2.3.3, 3rd paragraph, 2nd sentence
It was noted that we have made lots of changes around role and roleType so this issue should now be fixed by the changes made to resolve issue 1018.  It was agreed to close this issue with the official resolution of 'close will not fix', although changes to respond to the issue have been made under a different issue. 

Agreed to Close WONTFIX

NEW ACTION: Mark 10021 as Close WONTFIX

Issue 1038 Section 3 is Incomplete
Add an example? 
There is a syntactically complete example in section 2.5.2.3.  After some debate it was agreed to remove section 3 altogether from the specification.  It is intended to have a primer which will have a number examples and it is currently planned to issue this as a working group note. 

Agreed to Resolved fixed (by removing section 3)

NEW ACTION: Mark 1038 as Resolved Fixed

As a side issue it was agreed we should consider making the subsections of section two into their own sections.

NEW ACTION: Editor's to promote subsections of 2 up a level, so that each major subsection becomes a section.

Issue 1110 When a copy expression references a variable that has not been set, what should happen?
It was noted that an e-mail (http://lists.w3.org/Archives/Public/public-ws-chor/2005May/0036.html) has been sent by Gary on this issue and so it was agreed to put it on the agenda to be discussed. 

Issue 1126 
It was agreed to close no action as no proposal has been received.  The semantics of work units have been clarified and there is a work around. 

Agreed to Close WONTFIX

NEW ACTION: Mark 1126 as Close WONTFIX


Issue 957 XHTML Compliance
Done already but needs checking in the final published version as this issue only affects the XML (XHTML actually) of the final published version.

Issue 1000 CDL 'ROLE' Functions
Done and in the editing log.

Issue 1008 Fault Handling
This issue is on the agenda to be discussed.

Issue 1026 Information model - Section 2.3.4, 3rd last paragraph, last sentence
Related to session instance topic but resolved as a result of issue 1001.

New ACTION: Martin to closeIssue 1026, referring to 1001. 

Issue 1035 Point of clarity - Section 2.5.2.3, 11th paragraph, 3rd bullet
This issue is in the editing log already as done by Nick.  Update Bugzilla to mark as resolved fixed

NEW ACTION: Steve(/Greg?) to update Bugzilla to mark 1035 as resolved fixed. 

Issue 1039 Section 8 is Incomplete
This issue on having a Conformance section is already on the agenda.

Issue 1053 Define the bindings between related concepts in WS-CDL and WS-Addressing
Not in the editing log, but we have a section on relationship with addressing and further description in the primer.  We do not want to be too specific about addressing so it was agreed to write a resolution to the issue pointing to section 4 on the relationship to addressing and also pointing to the primer for more practical examples of how to use in practice.

In passing it was noted that it might be good to have a coordination group primer which showed how addressing and the various other specifications being produced by the W3C Web Services Activity all fitted together.

The group agreed not to make any change to the specification.  Therefore close no action on the specification, but open an action for the group to suggest a coordination group primer.

Agreed to Close WONTFIX

NEW ACTION: Mark 1053 as Close WONTFIX
NEW ACTION: Steve to suggest a W3C Web Services Activity coordination group primer.

Issue 1075 Channel Types - attribute "usage" should be made an Element
Anders has made a specific proposal so it was agreed to address this as an agenda point during this face-to-face meeting.

WSDL 2.0
In passing it was noted that we have a reference dependency on WSDL 2.0.  We may need to strike this if WSDL proceeds to slowly.  When WS-Choreography becomes a Recommendation, WSDL 2.0 needs to be at least a Proposed Recommendation.  If we need to make a change then will need to move WSDL 2.0 to the relationship section. 

Exit criteria
========
Refer to Steve's e-mail dated 10th May 2005 with Gary's comments embedded. 
http://lists.w3.org/Archives/Public/public-ws-chor/2005May/0028.html

Jeff Mischkinsky: need to show that all the features are implementable.  People usually do this by inventing a set of test cases.  Change the words 'an example' to 'examples'.  State that these are complete choreographies not just fragments.  Therefore the group needs to produce a set of choreographies that demonstrate all the features.

Item number 2 does not need to be there for the specification and (but perhaps as an exit criteria for the group to close).  Do need to add an implementation page to the website.

NEW ACTION: Yves to produce an implementation page for the WS-Choreography website.

Jeff said that WSDL group were making their primer part of the Recommendation unlike this group.  Should we make the primer part of the Recommendation?  The group agreed no, and preferred a separate Working Group Note.

Item one is also already in the specification.

It was therefore agreed to remove the current items 1 and 2 from the exit criteria and instead to break down the current item three into points. 

Need distinct implementations - at least on separate platforms.  Need to cover more than one role. 

'Encoded by' meant that two different organisations get together to agree the choreography description and any changes to it.

Martin thinks that having two or more validating editors is not important for the exit criteria.  The more interesting part is to have different implementations of the run time for CDL performing the same choreography description and to show that they can interoperate.

3b  requires that two distinct endpoints can be generated by different implementations and can be shown to interoperate correctly. 

Steve projected his words for this part of the exit criteria and the group wordsmithed the exit criteria.  Steve recorded the result in real time.

Need to define the term 'platform'.  There was some discussion as to whether we should specify more than one vendor.  The general feeling was 'no' - specify different implementations only, although clearly there is more risk that the implementations are not so different if from the same vendor, and therefore this is not such a good test.  It was agreed to leave ambiguous for the present.

A definition for 'platform' was suggested as software stack.

Discussed the point about 'all features': there are some things left deliberately vague in the specification.  These are alignment, and coordination protocol, and the precise addressing mechanism.

It was noted that 'WSDL' group is producing a document that specifies test descriptions for all the MUSTs in the specification.

So change 3 to not require interoperability using alignment, coordination and a particular addressing scheme.  Validate by simulation or by real endpoint interoperability?  Could leave it to the implementers producing the implementations to agree the alignment and co-ordination protocols, and the addressing scheme, to be used in the interoperability tests.  Otherwise we may have to define conformance profiles in the specification. 

Steve: could leave open at this face-to-face and engage in discussions with the implementers.

Everyone present agreed we did want to exercise all the features and that we were happy to leave the decisions on alignment protocol, coordination protocol, and the addressing scheme to the implementers. 

NEW ACTION: SRT to craft exit criteria based on decisions at F2F to send to editors to fold in to document.

Jeff suggested adding a subsection to the specification on interoperability - actually agreed to transfer to the section 4 on relationships.

NEW ACTION: Editors to reword section on Relationships to be a section on Interoperability (same content)


Issues Discussion
==============  

Issue 1039 
----------------
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1039

Discussion of Tony's latest proposal on conformance 

Proposal is at: http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/att-0010/2005-06-06_Contribution_to_WS-Chor_on_Conformance_section.doc 

Martin Chapman: delete 8.1 title and first sentence. delete However in 2nd sentence 

This section defines the conformance criteria that can be claimed for various CDL artifacts. 

proposal: replace 1st para with the above 

PROPOSAL: Replace first paragraph in proposal with "This section defines the conformance criteria that can be claimed for various CDL artifacts." 

<Steve Ross-Talbot> SECONDED - SRT 

<SRT> OBJECTION: NONE\ 

<SRT> AMDENDMENT ONE - PASSED 

PROPOSAL jm: delete para 2 of 8.1 

SECOND: Tony 

AMENDMENT TWO - PASSED 

discussion about the diffs between the two 8.2 sections -- realization that there are no words in the spec that say that the schema is normative 

Yves Lafon: there is nothing that says that appendix A is normative 

Martin: real PROPOSAL -- add " where there is disagreement between schema fragments contained in the normative part of the specification and the full schmea in appendix <A..Z>, the schema contained in the appendix shall take precedence 

<SRT> PROPOSAL: In the event of any amiguity or differentiation between the CDL fragements the appendix with the full schmea takes precedence 

Martin: amend PROPSAL to add in new section 1.7 
... called Authoratative Schema 

Tony Fletcher: SECOND 

APPROVED - no objection 

NEW ACTION: Editors to add new secion 1.7, Authoratative Schema, with content as agreed at the F2F

SRT: PROPOSAL -- A CDL document is conformant if it conforms to the schema in appendix <X> AND the additional syntactic and semantic contstraints found in sections 2 .. <Interop -1> -- 
... replaces both possible 8.2 
... and the root element must be a package as defined in the WS-CDL namespace 

note: the above is still in a section called Conforming WS-CDL documents 
... this is AMENDMENT 3 

PASSED - no objection 

8.2 Processor Conformance 

Martin: argues that there is an "entity" that produces observable messages that conform to the message flows defined by the choreo def 

Jeff: there is no CDL processor per se -- only have CDL docs that are conformant 

Martin: a vendor that claims that they are a CDL vendor does more than parse CDL docs 

<Paul Downey> David Booth gave a presentation on the difference between lanugage and processor conformance to the WSDL WG: http://www.w3.org/2004/Talks/0304-dbooth-wsdl-lang-vs-proc/ 

Tony: 3 things that can have conformance criteria applied -- 

-- document 

-- something that takes in a document and validates it, 

-- something that produces a conformant CDL document 

Jeff: this is really all the same thing -- XML docs that conform to the CDL spec are valid 
... what about defining a "CDL endpoint" which is the entity that properly executes a role define in a choreography 
... refinement - A conformant CDL endpoint is an entity that correctly implements the observable behavior of a role in a CDL choreography 

Monica: do we need to say something about implementing ALL of the roles in a choreography? 

SRT: probably not, it will difficult/impossible to determine when/if ALL the roles are available -- you will get an exception if you try to use a role that is not available 

Jeff: PROPOSAL -- replace section 8.3 with CDL Endpoint Conformance with the above statement about CDL endpoint conformance 

Monica: how do you take care of the case that some of the roles are "permanently" AWOL -- i.e. you don't have a "complete" choreography 

Tony, SRT, Martin: you need to re-write your choreography becuase your "model" doesn't conform to reality -- we have defined the faults that will occur in that case 

SRT: SECOND 

PASSED - no objection 

SRT: PROPOSAL strike 8.4 -- subsumed by the above 

PASSED - no objection 

<SRT> <SUMMARY FROM STEVE> 

<SRT> PROPOSAL PRESENTED AT F2F June 2005: http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/att-0010/2005-06-06_Contribution_to_WS-Chor_on_Conformance_section.doc 

<SRT> AMENDMENT ONE - Change first paragraph of section 8.1 to the following text: 

<SRT> "This section defines the conformance criteria that can be claimed for various CDL artifacts." 

<SRT> AMENDMENT ONE - ADOPTED 

<SRT> AMENDMENT TWO - Remove paragraph two from section 8.1 

<SRT> AMENDMENT TWO - ADOPTED 

<SRT> AMENDMENT THREE - Replace both of section 8.2 with: 

<SRT> "A CDL document conforms to the schema in appendix X plus the additional syntactic and semantic constraints defined in the specification text in sections 2 through Z. The root element must be package as defined in WS-CDL namespace. " 

<SRT> AMENDMENT THREE - ADOPTED 

<SRT> AMENDMENT FOUR - Replace section 8.3 with the following: 

<SRT> "A conformant CDL endpoint is an entity that correctly implements the observable behavior of a role defined in a CDL choreography" 

<SRT> Replace section heading with "Endpoint conformance" 

<SRT> AMENDMENT FOUR - ADOPTED 

<SRT> AMENDMENT FIVE - Remove section 8.4 

<SRT> AMENDMENT FIVE - ADOPTED 


PROPOSAL: change status of 1039 to EDITORIAL and mark it as resolved later 

PASSED - no objection 

NEW ACTION: Mark Issue 1039 as editorial/resolved later, with the actual text defined in F2F minutes.

issue 1110
---------------
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1110
----------------

 http://lists.w3.org/Archives/Public/public-ws-chor/2005Apr/0024.html 

<MChapman> Variables used in any action except for a blocking workunit or used 

<MChapman> in the CDL function isVariableAvailable the result is undefined 

<MChapman> insert: uninitialised 

<MChapman> Variables used in any action except for a blocking workunit or used 

<MChapman> in the CDL function isVariableAvailable the result is undefined 

SRT: PROPSAL adopt with insertion of unitialized into 2.4.2 

<MChapman> uninitialised Variables used in any action except for a blocking workunit or used 

<MChapman> in the CDL function isVariableAvailable the result is undefined 

SRT: PROPOSAL adopt with insertion of uninitialzied into section 2.4.2 

PASSED - no objections -- change status to EDITORIAL and mark resolve later 

NEW ACTION: Mark issue 1110 as editorial/resolved later with wording agreed at F2F

Issue 1008
---------------

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1008 

http://lists.w3.org/Archives/Public/public-ws-chor/2005May/0049.html 

 2 possibilities: 1. no change 2. adopt above 

time box - 10 minutes 

Nick Kavantzas: can i have an op with 2 fault names f1, f2 and have the same tag -- if i use op in an interact, can the client tell (in an interop fashion) to disambiguate 
... current spec has limitation that you can't tell - unless both sides use the same engine 
... leave as is, since there's not much that can be done 

Charlton: don't have a consistent way to handle fault types between wsdl 1.1 and wsdl 2.0 - with addition of fault name you can use info type and fault name to disambiguate faults in 1.1, but also gives you consistency between 1.1 and 2.0. In addition you can reduce informationType proliferation for faults which share the same type.

Gary Brown: nick's argument is based upon there not being wire protocol support [for this] currently -- [however,] wsif does support this 
... and this feature would make it easier to support forward generation 

straw poll: close no change: 0 adopt proposal: 4 abstain: 4 

any objections to adopting the proposal: NONE 

PROPOSAL: adopt the proposal -- PASSED - and update bugzilla and mark as ed and resolve later 

NEW ACTION: Mark Issue 1008 as Editorial/Resolved later and incorporate the proposal agreed at the F2F


Issue <xxx> (1493)
---------------
http://lists.w3.org/Archives/Public/public-ws-chor/2005May/0075.html 
http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/0011.html 

NEW ACTION: martin to enter issue <xxx> in bugzilla 
DONE and issue number for this is 1493
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1493


Martin: seems like the choice is to add more semantics or simplify to the proposal 

NEW ACTION: Tony to check with peter to see if there is a reason why gary's proposal might be flawed 

NEW ACTION: martin to put on agenda confirmation vote to adopt Gary's proposal unless Tony gets some information from Peter that the group beieves means gary's proposal should be adopted 

--- confirmation vote to occur tomorrow 

issue 1116 
----------------

 http://www.w3.org/Bugs/Public/show_bug.cgi?id=1166 

Martin: big question is does CDL ver 1 have to have this to support use cases in the reqs? 

shouts of YES 

Martin: we should look at what BPEL has done 

delay this one for tomorrow 

issue 1459 
----------------

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1459 

consensus seems to be to handle the issue similar to the way that bpel did .

Proposal to add: "An isolated choreography cannot directly or indirectly perform another isolated
choreography."

Proposal 


Proposal accepted change to ED and resolve l8r -- PASSED - no objection 

NEW ACTION: Mark 1459 as editorial/resolved later with text agreed at F2F.


<end day 1>


Formalisms and CDL
================

Kohei Honda: Presenting "Programming interaction with Types" 

http://www.w3.org/2002/ws/chor/5/06/F2FJune14.pdf

Kohei: pi-calculus as foundation 
... pi-calculus as basis for CDL and what this means for the enterprise 

Nick asks Kohei to explain binding for CDL Calculus 

Can we scope variables, e.g., at different levels, as we do in CDL? 

Kohei explains several examples of CDL calculus using Buyer->Seller->Shipper scenario 

Kohei describes that causality on actions should always be local 

Kohei elaborates on connectedness principles 

Kohei blazes through a number of concepts 

Nick and Kohei in discussion 

Discussion on EPP-GVE (End-Point Projection - Global ViEpoint)

Autogeneration view - have endpoint here, automatically generate inverse - can generate one of a number of possible compatible patterns 

Session types provide the EPP-GVE contract 

Martin: Couple of discussion points to review after Kohei's talk 
... Will Kohei's action plan include analysis by this group? 

SRT: Yes - it w/b very useful to have Kohei's paper published as a Working Note 

Kohei: Understanding on requirement for software tools for CDL - what is not so clear to us is what tools vendors would think is necessary 

SRT: W3C critical success factors 
... Analysis - roundtripping between global and local calculus 

The Group thanked Kohei for his presentation and work, and requested they consider publishing as Working Group Note through this work group.


Issues Discussion (cont)
===================

Issue 1493 
----------------
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1493

Ordering activities in finalizers 

Note: Peter is in attendence 

Issue is that spec said nothing about the ordering in a finalizer block. 

Peter Furniss has clarified that any ordering is possible, but  cant remember the rationale for goruping the way it is

Martin: Says isn't it better to adopt the proposal from Gary, and use normal structing concepts in cdl (sequence/parallel). 

http://lists.w3.org/Archives/Public/public-wd-chor/2005Jun/0007/html 

PROPOSAL ABOVE IS ADOPTED 

NEW ACTION: Mark 1493 as editorial/resolved later with propsal agreed at F2F


Issue 1003
----------------
Issue: http://www.w3.org/Bugs/Public/show_bug.cgi?id=1003 

Nick asks for clarification of "shared" in thid proposal in the bugzilla entry under item (3) in comment. 

Nick: Can I have two receivers at the participant that use the same channel? 

If they are in the same session but different op names then this is okay? 

Kohei: If the op names are diff then it is okay 
... If we use a session we can assume that their are separate instances of channels and/or the same operation. 

Latest: http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/0001.html 

http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/0001.html 

Nick: If in item three we remove "client side" then it is okay for me. 

Gary: But the client side is important because you cannot share the server side of a channel instance only client side of a channel instance. 

Kohei: Sharing is the output capability in the pi-calculus. Inputting capability can only be owned by one participant. 

Item (1) of the proposal is okay. 

Item (1) clarification - once means a single interaction. 

Item (2) is about owning participant 

Definition: Once applied to a channel type means that a channel instance of that type can be used once and once only to interact and nothing more. 
... Once applied to a channel type means that a channel instance of that type can be used once and once only to interact and nothing more. 
... Once applied to a channel type means that a channel instance of that type can be used once and once only for a single interaction and nothing more. 
... Once applied to a channel type means that a channel instance of that type can be used once and once only for at most one interaction and nothing more. 
... Once applied to a channel type means that a channel instance of that type can be used once and once only for at most one interaction or can be passed to another role. When a channel instance of this type is passed the passer of the channel instance reliquishes (for outputting only) control of that instance and passes control to the receiver. 

Add: A channel instance of this type MUST NOT be passed to more than one receiver at any one time. 

Definition: Distinct applied to a channel type means that a channel instance of that type can can be used multiple times by a participant within multiple interactions. When a channel instance of this type is passed the passer of the channel instance relinquishes (for outputting only) control of that instance passing the control to the receiver. A channel instance of this type MUST NOT be passed to more than one receiver. 
... Shared applied to a channel type means that a channel instance of that type can be used multiple times by multiple participants within multiple interactions. When a channel instance of this type is passed the participant passing the channel instance does not reliquish control. 

These (above) are the working definitions agreed on my the WG 

<charlton> +1 to default is distinct 

The default behavior for channels is "distict" 

Note: For distinct and shared we need to add text about operation interference. 

NEW ACTION: issue 1003, Gary to draft proposal text based on definitions agreed at F2f


Issue 1101
----------------
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1101 

The issue was clarity around use of completion and coordination protocol. 

RESOLVED WON'T FIX - resolved based on coordination=true 

NEW ACTION: Mark 1101 as Closed Resolved WONTFIX based on agreement at F2F


New agenda item: exploring current session mechanism with session ideas from cdl calculua
-------------------------
Based on Kohei's presentation we discussed whether there was any mis match between sessions as recently adopted (http://lists.w3.org/Archives/Public/public-ws-chor/2005Jun/0000.html ) and session in Kohei's work, as maybe we have more work to do.

Does the new correlation/identity mechanism provide the necessary features for session types in WS-CDL? 

Gary thinks that the current mechanism do provide what they need. Even though channel instances may get reused they would be renamed and so act as a separate/new channel. 

Nick points out that the "new" attribute implies some mechanism that enables channel instances to be created and uniquely named (distinct) without specifiying an operator to do this. Therefore the implementor is responsible for enforcing channel creation and unique naming of channel instances. 

The conclusion is that the status quo provides all of the necessary facailities through "new" and "scoping of sub choreos" and the "correlation of identity" 

NEW ACTION: Kohei - write how to make the implicit session model explicit (issue 1496).


Issue 1075
-----------------
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1075 
http://lists.w3.org/Archives/Public/public-ws-chor/2005Apr/att-0021/WSCDL_Comments1075_Attr2Elem.pdf 

WG feels that the existing extensibility mechanism is sufficient to meet the requirements in this proposal. 

New attributes and elements may be added based on the current schema defn. 

Agreed to Close WONTFIX

NEW ACTION: Mark 1075 as Closed WONTFIX based on decision at the F2f 


Planning
=======
MC sketches out state of play 

MC: Once we have finished all of our issues we can publish a new spec and go to CR. 
... We can then do minor changes that do not change the architecture of WS-CDL. 
... When exit-criteria-is-met = true 
... Go to recommendation (R) 

Martin: We're three issues away from being finished with our issues. 

Issues to finish for CR: 

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1496 

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1166 

http://www.w3.org/Bugs/Public/show_bug.cgi?id=1003 

Of these only 1166 has any additional semantics required. The other two do not require further semantics, and only exact wording required based on agreed text/defintions. 

Cut-off date for all of these items is 22nd July. After that time the group will make a decision to move to CR or not. 

In parallel we need to work on two other work items: 

(1) Primer 

(2) Test examples for exit criteria 

(3) Working note for formal underpinings 

Primer by end of Oct 2005 

Example for exit criteria for mid August 2005 

The will be a meeting at Oracle HQ on 29th June to flesh out example.
Martin and Charlton will attend, others are welcome.

Next F2F - Sophia Antipolis maybe - dates to be confirmed next week, but close to bpel f2f if in europe.

NEW ACTION: Chairs Confirm dates for F2F SRT 

Issue 1166
----------------

Further discussion of 1166: 
http://www.w3.org/Bugs/Public/show_bug.cgi?id=1166

Tony: WS-BPEL issue 6.2 - about Flow - looking at same problem as 1166? 
Gary: Has nothing to do with it. 
Nick: Actually issue 147 - for each - is related to this 
Nick: 6.2 is complete condition in a Flow 
Tony: Other assertion - problem of what you do with dangling threads 
Gary: Issue 1166 - in this scenario we do not have a complete condition 
Tony: Our description can use extension points to bring in xsd, xpath, xquery which s/b adequate 
SRT: Entirely possible and practical to have the manipulation be silent 
Tony: What is the fundamental problem 
SRT: It is in the successful completion where we have dangling threads 
SRT: Where we have successful completion and we don't have dangling threads - we've solved that problem 
Nick: Trigger complete condition in other chor? That would complete dangling threads therein 
Gary: If chor finishes and we have dangling threads, there is an assumption that dangling threads be dealt with implicitly 
Nick: Can we change the name from asynchronous to something else - say blocking? 
Martin: Attribute on perform which basically says that allows us to immediately return and carry on with the rest of the activities 
Be it "async", "blocking", etc. 
Gary: Async join-on-sub-chor issue 

Martin: Where do we put this condition? 

Gary: This would be a when/guard condition - you'd have a WorkUnit repetition - spawn off with other WorkUnit with guard condition - isComplete(name, [chorId], ...) * - use same mech for initial look to check isComplete() - or any sub-chor with name x, you wait for 

Gary: And autocomplete behaviour - what if parent is terminated, causes all subs to be killed 
Martin: So, if don't complete chor here (don't finish), dangling threads continue to run, this is a design issue t/b described in the primer 

Kohei: So, after completion, they can still be running? 

Gary: No - no completion handler - after completion, they just terminate 

Martin: Has chor finished is the behaviour to capture.... Choreography lifeline.... 

Gary: Do we want another test for whether a chor completed successfully? 

SRT: Maybe two - one for success and one for failure>? 

Martin: getChoreoStatus(name, [id]) 

Gary: hasChoreographyCompleted() used to hedge off completion - testing chor status maybe useful, but maybemore useful to do check that related chors h/b completed.... 

Nick: This shortcut isn't consistent with how we do finalize 

 Martin: Happy to have Gary + Nick to go off and think this over 
 Nick: Two more items w.r.t. 1166.... 

NEW ACTION: Issue 1166, Gary to rewrite proposal to this to capture editorial changes 

AOB
====
Martin: Let's bring meeting to a close 

Thank you Monica and Sun for hosting this meeting 

Thank you Kohei, Nobuko, Marco for participating at this F2F 

Thank you Tony for your participation 

Martin: Next F2F - primer, examples 
Thank the chairs for a great meeting 

Summary of New Action Items
========================
NEW ACTION: Mark 1004 as Close WONTFIX
NEW ACTION: Mark 1005 as Close WONTFIX
NEW ACTION: Mark 1021 as Close WONTFIX
NEW ACTION: Mark 1038 as Resolved FIXED
NEW ACTION: Editor's to promote subsections of 2 up a level, so that each major subsection becomes a section.
NEW ACTION: Mark 1126 as Close WONTFIX
NEW ACTION: Martin to close Issue 1026, referring to 1001. 
NEW ACTION: Steve(/Greg?) to update Bugzilla to mark 1035 as Resolved FIXED.
NEW ACTION: Mark 1053 as Close WONTFIX
NEW ACTION: Steve to suggest a W3C Web Services Activity coordination group primer.
NEW ACTION: Yves to produce an implementation page for the WS-Choreography website.
NEW ACTION: SRT to craft exit criteria based on decisions at F2F to send to editors to fold in to document.
NEW ACTION: Editors to reword section on Relationships to be a section on Interoperability (same content)
NEW ACTION: Editors to add new secion 1.7, Authoratative Schema, with content as agreed at the F2F
NEW ACTION: Mark Issue 1039 as editorial/resolved later, with the actual text defined in F2F minutes.
NEW ACTION: Mark issue 1110 as editorial/resolved later with wording agreed at F2F
NEW ACTION: Mark Issue 1008 as Editorial/Resolved later and incorporate the proposal agreed at the F2F
NEW ACTION: Martin to enter issue <xxx> in bugzilla 
        DONE and issue number for this is 1493
NEW ACTION: Tony to check with 
peter to see if there is a reason why Gary's proposal might be flawed 
NEW ACTION: Martin to put on agenda confirmation vote to adopt Gary's proposal unless Tony gets some information from Peter that the group beieves means Gary's proposal should be adopted 
NEW ACTION: Mark 1459 as editorial/resolved later with text agreed at F2F.
NEW ACTION: Mark 1493 as editorial/resolved later with proposal agreed at F2F
NEW ACTION: Issue 1003, Gary to draft proposal text based on definitions agreed at F2f
NEW ACTION: Mark 1101 as Close resolved WONTFIX based on agreement at F2F
NEW ACTION: Kohei - write how to make the implicit session model explicit (issue 1496).
NEW ACTION: Mark 1075 as Close WONTFIX based on decision at the F2f 
NEW ACTION: Chairs Confirm dates for F2F SRT 
NEW ACTION: Issue 1166, Gary to rewrite proposal to this to capture editorial changes 

<end day 2>