Raw minutes from 2007-Oct-11 telcon

["Charlton Barreto" <barreto@adobe.com>]

Raw minutes with some cleanup - please let me know if you have any questions or corrections.

 

Anish, can you confirm the roll call for the telcon? I may have missed someone.

 

-Charlton.

OASIS SCA-BPEL TC 
11 Oct 2007

Dial-in:
1-888-967-2253
+1-650-607-2253
+61-2-8817-6100
+44-118-924-9000

Meeting ID: 877770
Meeting password: SCABPEL (7222735) 

[8.11] charlton: roll call
    Mark Ford, Active Endpoints, Inc.
    Charlton Barreto, Adobe Systems, Inc.
    Michael Rowley, BEA Systems, Inc.
    Mike Edwards, IBM
    Dieter Koenig, IBM
    Martin Chapman, Oracle Corporation
    Khanderao Kand, Oracle Corporation
    Anish Karmarkar, Oracle Corporation
    Ashok Malhotra, Oracle Corporation
    Alex Yiu, Oracle Corporation
    Najeeb Andrabi, TIBCO Software Inc.
    Danny van der Rijn, TIBCO Software Inc.
    
    [Anish: Anyone I have missed?]
    
    [late entries]
    Simon Moser, IBM
    
[8:11] charlton: scribe: Charlton

[8:11] scribe: agenda review

[8:11] scribe: topic: agenda review

[8:11] scribe: no comments on agenda

[8:11] anish: agenda: http://www.oasis-open.org/apps/org/workgroup/sca-bpel/event.php?event_id=16223

[8:12] scribe: topic: discussion of previous meeting's minutes

[8:12] scribe: no objections

[8:13] scribe: previous meeting's minutes approved

[8:13] scribe: topic: Administrivia (agenda item 6)

[8:14] scribe: topic: agenda item 6a

[8:14] scribe: a: editors have read access to JIRA?

[8:14] scribe: topic: agenda item 6b

[8:14] scribe: b: editors have write access to JIRA?

[8:14] scribe: yes to both

[8:14] scribe: topic: agenda item 6c

[8:14] scribe: c: approve Charlton to join editors list

[8:15] scribe: Jeff motions to add Charlton as editor, Najeeb seconds

[8:15] scribe: no objections, motion passes

[8:14] scribe: topic: agenda item 6d

[8:15] scribe: d: Issue process

[8:17] scribe: Martin: in proposal agreed to in slide deck, agreed that there are two substates of resolution - 1 resolve later, 2 resolve fixed. 1: resolve-later is when TC agrees resolutions and assigns AI to editors, editors apply solution in ctee draft, then can move to resolve-fixed

[8:17] scribe: Alex: Can we move a resolve-fixed back to resolve-later?

[8:17] scribe: Martin: Let's give it a go

[8:19] scribe: Martin: I was not able to un-resolve an issue - is our workflow supported? 

[8:19] scribe: Anish: Separate status to reflect resolve-later? 

[8:21] scribe: Alex: Editors to add comment to issue to indicate that it is fixed?

[8:21] scribe: Martin: Once an issue is marked resolved, no way to undo that resolution.

[8:21] scribe: Ashok: Resolved makes an issue basically closed. It can be reopened as a different issue.

[8:23] scribe: Anish: Martin to take an AI to address this process in JIRA. 

[8:24] scribe: Action: Martin to address the resolve-later/resolve-fixed process in JIRA.

[8:25] scribe: topic: Specification Development (agenda item 7)

[8:25] scribe: Dieter provides update from spec editors

[8:27] scribe: (Issue #7)

[8:28] scribe: Dieter: Word views - markup and w/o markup - markup not lost in spec, Dieter has markup in different view

[8:29] scribe: Alex: When editor applies resolution to issue, may want to add comment and link to document uploaded to OASIS

[8:29] scribe: Alex: to the Issue in JIRA

[8:29] scribe: Dieter: I will do that

[8:30] scribe: Michael: In addition, can add revision number to the editors' draft to the issue resolution

[8:31] scribe: Topic: New issues (agenda item 8)

[8:33] scribe: Topic: Issue #8 (agenda item 8a)

[8:32] scribe: Ivana: Steering ctee s/agree on version number to be used on all SCA specs

[8:32] scribe: Jeff: Move to accept Ivana's resolution

[8:32] scribe: Alex: Seconds

[8:32] Dieter Koenig: I added the comment "Fixed in working draft 3 - http://www.oasis-open.org/apps/org/workgroup/sca-bpel/document.php?document_id=25646"; to issues BPEL-4 and BPEL-7

[8:32] scribe: No objections, issue accepted

[8:34] scribe: Topic: Issue #9 (agenda item 8b)

[8:35] scribe: Dieter: SCA-Assembly, BPEL extensions, BPEL data type(s) - need to separate namespaces

[8:35] scribe: Anish: Clarification questions?

[8:35] scribe: Motion to approve?

[8:35] scribe: Dieter: Move to open Issue #9

[8:36] scribe: Alex: Seconds

[8:36] scribe: s/Move to accept Ivana's resolution/Move to open Ivana's issue (Issue #8)/

[8:36] scribe: Topic: Issue Discussion (agenda item 9)

[8:36] scribe: Danny r u there?

[8:37] scribe: Topic: Issue #6 (agenda item 9a)

[8:38] scribe: Danny not available to discuss Issue #6

[8:38] scribe: Alex: Ready to complete proposal with text to address Issue #6, to hopefully close next week

[8:38] scribe: Anish: Any disagreement with Alex's position?

[8:38] scribe: Michael: Ordering w/b document ordering

[8:38] scribe: Action: Alex to generate proposal Issue #6

[8:39] scribe: Topic: Issue #3 (agenda item 9b)

[8:39] scribe: Najeeb to summarise

[8:41] scribe: Najeeb: Message that cannot be correlated cannot throw an exception. Alex suggested that this s/b fixed. After going through spec, if silent about correlation mismatch, but BPEL correlation exception w/b thrown on invoke. Particular correlation would fail. Now, for conversation, if someone puts SCA conversations in, and uses WSA to correlate, and that part passes, but BPEL part fails. What does BPEL send back? BPEL correlation exception, but other conversation correlation does not?

[8:42] scribe: Michael: This gets us back to a receive case - where this is handled with a timeout

[8:43] scribe: Najeeb: In one case yes, but in other case - where corr id is sent back to you, but not for corr you created - it is an invoke - the synchronised case. 

[8:43] scribe: Dieter: For response to invoke, don't need correlation id.

[8:43] scribe: Alex is on the queue

[8:43] anish i see u alex

[8:44] scribe: Najeeb: In BPEL spec - one case where correlation false is sent back - in case of invoke

[8:45] Najeeb Andrabi: When a bpel:correlationViolation is thrown by an <invoke> activity because of a violation on the response of a request/response operation, the response MUST be received before the bpel:correlationViolation is thrown. In all other cases of bpel:correlationViolation, the message that causes the fault MUST NOT be sent or received.

[8:46] scribe: Alex: Clarification - corr violation fault is never on the wire. In case of invoke, if mismatch be payload data and corr data for invoke, corr violation fault will be caught by the process instance only. Conversation will never see it. Secondly, in order for ws client to bpel process to be notified by some fault in correlation, the underlying protocol must be known, otherwise it is difficult for any spec to indicate what action must be taken.

[8:47] scribe: Mike is on the queue

[8:47] scribe: Michael is on the queue

[8:49] scribe: Michael: Related to conv violation fault in Assembly spec. Issue lurking here, but not certain what it is. Use of SCA's conversation mech when not using engine (BPEL) managed corr, can get SCA violation fault when use conv id or cont to use conv id when expired, couple of instances in the spec. Suspect that concern of getting comb of SCA conv mech and BPEL corr mech falls out. Works by implication w/o add'l text.

[8:50] scribe: Najeeb: Fundamental point - timeout s/n be way of telling client that corr failed

[8:51] scribe: Najeeb: Case of one BPEL process interacting with another

[8:51] scribe: Michael: How do you tell it is invalid rather than a valid value that hasn't become valid yet

[8:51] scribe: Alex on queue

[8:53] scribe: Alex: We tried to hash out these points in the email thread - corr set is not valid yet, where incoming message with corr where receive has not be initiated yet.

[8:53] Mike Edwards: can't the service end detect that there is a problem and raise a fault ??

[8:53] Mike Edwards: there are all kinds of ways of transmitting a fault....

[8:53] scribe: Alex: New protocol agreement to add a third fault type address this

[8:54] scribe: Alex: Without this agreement, it is difficult to determine why the violation occurred

[8:55] scribe: Najeeb: Unacceptable to use timeout to reflect corr violation. And SCA can have successful conv but component level corr can fail. Telling a client that corr has failed based on timeouts is bad design.

[8:55] scribe: Michael: Why is it that intro of new fault implies new protocol?

[8:56] scribe: Alex: Not necessarily intro new protocol, can be additional aspect to existing protocol. Look at WS-BA - assume client and BPEL process using WS-C and WS-BA

[8:56] scribe: Alex: In error case need to align with underlying WS-C/WS-BA protocol

[8:57] scribe: Alex: Correlation failure - message detection not defined in WS-C/WS-BA, need to explore this in order for this to work.

[8:58] scribe: Mike Edwards: If client talking to service, if provide bad args, expect to receive fault from service - 'you've provided invalid data in request'

[8:58] scribe: Alex: This is one of the cases that we can detect. But the situ is more complex w.r.t. ME between consumer and provider - arrival of message and enablement of incoming message activity (receive).

[9:00] Mike Edwards: +1 - I agree

[9:00] anish: i agree as well

[9:00] anish: the part about application fault v. infrastructurual fault

[9:00] scribe: Michael: Don't believe we have to be concerned about how interact with WS-C and WS-BA. If can determine that message is never to be accepted/valid, s/b able to throw a fault from the BPEL process instance. In presence of coord protocol, s/b able to handle app double faults. If can ever determine that incoming message can never be accepted, believe that there are cases where this can be done. 

[9:01] scribe: Alex: Discuss further in email. But want Michael Rowley how to send fault back to client if MEP is one-way/in-only

[9:01] scribe: Alex: (without fault)

[9:01] scribe: Michael: Agree in that case, cannot do it

[9:02] scribe: Anish: Encourage taking this to email toward making a proposal

[9:02] scribe: Topic: Quick discussion over Issues #1, 2 and 9 (agenda items 9d, 9e, 9f)

[9:02] scribe: Anish: Martin to make a proposal on Issue #1?

[9:03] scribe: Anish: I'm to make a proposal on Issue #2

[9:03] scribe: Anish: Dieter included proposal with Issue #9

[9:03] scribe: Topic: Any other business (agenda item 10)

[9:03] scribe: Any stragglers?

[9:03] scribe: Simon Moser (IBM) - on roll

[9:04] scribe: Anish: Adjourn meeting until next week (Oct 18)