Minutes WS Choreography WG conference call 12th October 2004
W3C WS-Choreography group
Teleconference Notes for 12 October 2004
Tony Fletcher (Choreology), Steve Ross-Talbot (Co-Chair - Enigmatec), Gary Brown (Enigmatec), Greg Ritzinger (Novell), Martin Chapman (Co-Chair - Oracle), Nickolas Kavantzas (Oracle), Monica Martin (Sun), Charlton Barreto (webMethods - IRC only), Yves Lafon (W3C staff).
irc log at http://www.w3.org/2004/10/12-ws-chor-irc
Scribe: Tony agreed to scribe.
The minutes for the September 2004 face-to-face meeting were approved:
The minutes for the teleconference held 5th October 2004 were approved:
ACTION 1: Take the TWIST example we're working on, add part of the flow what can go wrong, see what happens when things go wrong. We can see in front of us the choreography being executed, what we do today with current language, what we can do with transaction. Steve is starting to add exception handling. Nick and Steve have corresponded on "perform" which revealed some errors in the specification to be corrected.
ACTION 2: Chairs to sort out examples work.
ACTION 3: Items related to time: Update the CDL one Charlton (has deadline reached, Nick to clarify the semantic on the time functions).
DONE - Charlton has sent a proposal on deadline time-outs.
ACTION 4: on Martin to propose text on clocks (Issue 885).
IN PROGRESS - to be done.
ACTION 5: Nick to clarify the text wrt NOT and OR operators in guard conditions.
IN PROGRESS - Nick said that had not done anything specific on OR or NOT - not sure what is expected. Steve explained that the action is about the order the guard condition should be evaluated. Wait for all variables to have a value or only those on one side of the OR expression. Nick replied that maybe this was an issue rather than an action. He confirmed that currently the specification requires all the variables to be available before a guard condition is evaluated, but agreed that it would be possible to allow a short-circuit once the overall value is determinable. (For instance, if OR then when one side=true and the whole expression =true whatever the values of the other variables).
ACTION 6: on the chairs to discuss part 2 of the specification
NO PROGRESS - at face-to-face talked about a second part of the specification that covered best practice, implementation hints, warnings on race conditions, live locks, deadlocks and so forth.
ACTION 7: Gary to demonstrate how an out-only pattern might be used in CDL (without any changes).
DONE - Gary has made a proposal.
ACTION 8: Bob/Choreology to come back with a definitive proposal with editing instructions on PROPOSAL ONE
ACTION 9: above question (is finalizer role bound?) needs to be answered in a proposal on PROPOSAL FOUR
ACTION 10: Write up concrete proposal on PROPOSAL FIVE
All 3 above actions are IN PROGRESS
Discussion: Martin requested that the Choreology proposal split out solutions to the separate proposals. Tony replied that Choreology may well provide an overall paper to show how everything fits together but would try to put in separate proposals, if this made sense i.e. the proposals could stand on their own, but Tony expressed concern that this was not case and our proposals would actually be linked. Tony said that we hoped to have something available in two to three weeks' time.
ACTION 11: Martin puts text for issue 885
It was agreed this a repeat of action 4 above.
ACTION 12: put issue 870 in the agenda for next con-call (time-boxed)
DONE (see agenda)
ACTION 13: propose a list of standard faults
REMOVE - this is already an issue - remove as action.
ACTION 14: editors to insert at end of section 2.2.4 the following:
Nick said cannot trace where added so not sure how to describe - no new attribute under the Description element.
ACTION 15: Martin to produce UML model of WS-CDL for the editors to include in spec.
REMOVE - this is an issue. Martin said that he would do but not until towards the end. Remove as an action.
ACTION 16: Monica to check if 687 has been resolved in latest draft.
OPEN - this issue is down to Monica who was not available at the time - leave open - she has sent in text for issue 687.
ACTION 17: SRT to check if 691 has been resolved.
DONE - Steve has sent in a proposal to resolve this issue.
Steve said that they were two further actions that are not mentioned on the agenda. Charlton has agreed to make proposals on WSDL, and on bindings.
Issues and proposals
The current reactive mechanism introduces implicit links between concurrent activities that compromises simplicity and makes model verification difficult.
Gary has provided an example of why it is complex. Makes model checking difficult. See Gary's e-mail link above - may need input from Robin/Nobuko. Nick said the problem is endemic. Gary thinks may be too late to support for version one but worth putting in a warning about synchronisation problems.
ACTION - send Bugzilla Issue 870 packaged up to Nobuko/Honda see if they can help. Steve to do.
Gary has outlined how this could work in CDL. One way out only. Please refer to his PDF and e-mail with a WSDL 2 definition in (links above).
All three operations are associated with the “to” role, one is in / out, and the other two are out only. All are on the same Channel. There was some discussion on the use of Channel in this manner. It was noted that WSDL has not defined a binding for out only at present. What Gary has tried to show is that we do not have to change CDL to cope with this aspect of WSDL 2.0. We would need proposal to restrict CDL to disallow what Gary has proposed. Martin thinks that the current text does preclude this already. Gary and Steve think we should allow this usage.
ACTION: Gary/Steve to propose text change to specification, if any.
Gary referred to his email on fault messages (link above).
Has WSD with fault messages and shows how CDL can show a choice between response message or one of the two fault messages.
Nick said that he has just sent a big paper on exceptions (http://lists.w3.org/Archives/Public/public-ws-chor/2004Oct/0026.html). He thinks Gary's ideas are included. This is issue 565.
ACTION: the chairs to consider this topic again next week
Issue 691 - locally defined variables:
See proposal from Steve (http://lists.w3.org/Archives/Public/public-ws-chor/2004Oct/0021.html)
Steve has checked the editors text and is suggesting that roleType attribute on the variable element becomes zero or more.
The intent is to be able to list the roles at which the variable is available for use.
Martin stated that each variable of a given type would have different instances at each role.
This proposal was adopted.
ACTION: Steve to tidy up proposal and send in (including Schema changes) to the editors.
Issue 913: Globalised trigger - Clarification on workunit guard projection:
Gary sent an e-mail on 7th of October about work unit with a guard that has a variable that is only available at one side (http://lists.w3.org/Archives/Public/public-ws-chor/2004Oct/0022.html). Means guarded choice at one end, unguarded choice at the other. Nick said yes can have unbalanced work units. Becomes unbalanced when you do projections on to each end.
Steve said that we should keep the issue but reassign to a primer, or part 2. Nick commented that type system may help here.
Have to accept that there are good and bad ways of using CDL and write a best practice/examples to show the problems and how to handle them.
ACTION: Greg to re-assign issue 913 to primer and / or Part 2
Nick said he has added in time-out functions and explained at the recent face-to-face meeting how they can be used to meet the time out challenge Tony set, which could not be met before.
Yves announced that the new working draft has been published (http://www.w3.org/TR/2004/WD-ws-cdl-10-20041012/).
There being no other business and the meeting being out of time the meeting was closed.