Agenda_20040317.htm
|
Chairs: |
|
|
Oracle | |
|
Enigmatec | |
|
|
|
|
W3C Staff Contacts |
|
|
| |
|
|
Members:
|
Novell | |
|
Oracle Corporation | |
|
Intalio Inc. | |
|
Choreology Ltd | |
|
Commerce One | |
|
SeeBeyond Technology Corporation | |
|
Cisco Systems Inc | |
|
SAP AG | |
|
IONA |
Raw irc log at: http://www.w3.org/2004/03/16-ws-chor-irc
David Burdett volunteered to scribe.
The following is a listof recent scribes (in order): Kevin Liu, Jeff Mischkinsky, Monica Martin, Nick Kavantzas, Greg Ritzinger, David Burdett, Tony Fletcher, Monica Martin, Ugo Corda, Mayilraj Krishnan, Ravi Byakod , Martin Chapman, Steve Ross-Talbot, Monica Martin, Nick Kavantzas, Ed Peters, Anthony Fletcher, Jeff Michkinsky, Dinesh Shahane, Greg Ritzinger, Ed Peters, David Burdett,Ivana Trickovic, Ugo Corda, Assaf Arkin, Monica Martin, Carol McDonald, Nick Kavantzas, Tony Fletcher, Mayilraj Krishnan, Francis McCabe, Jeff Mischkinsky, David Burdett, John Dart, Monica Martin,Tony Fletcher, Jim Hendler, Kevin Liu, TonyFletcher, Jon Dart, DavidBurdett,Ed Peters, Greg Ritzinger, Monica Martin, Len Greski, Jean-JacquesDubray,Monica Martin, Mayilraj Krishnan, Francis McCabe, Michael Champion,AbbieBarbir, David Burdett, Jon Dart, Carol McDonald,Yaron Goland, Leon Greski,Ed Peters, Greg Ritzinger, Daniel Austin, PeterFurniss, Jim Hendler
Minutes approved.
ACTION: SRT will update bugzilla entries to reflect exceptions/error discussion. (0127, 0203, 0210, 0217, 0301, 0309) DONE
Banana calculus review - has been logged against bugzilla - done
ACTION : The chair will schedule an agenda item in a con call to discuss intermediate end-point language. DONE
On this Agenda
ACTION: Chairs to look into a far east meeting in sep/oct (0113, 0120, 0127, 0203, 0210, 0217, 0301, 0309, 0316) IN PROGRESS.
Far east meeting - discussed last week, review at end of action items
ACTION: Chairs to organise an issues conference call in which the regular conf call is used with members to look at the issue in bugzilla (0113, 0120, 0127, 0203, 0210, 0217, 0301, 0309, 0316) IN PROGRESS.
Martin - need a call to do triage of the list or schedule it in this call. Prefers the idea of a sub group to tidy things up. Suggests changing action. Talk later ...
ACTION: Chairs to raise a topic of Binding, Context etc. (0113, 0120, 0127, 0203, 0210, 0217, 0301, 0309) DONE
Binding context - Steve R anyone want to talk about it
David B. Wants to talk about how binding to WSDL would work.
Steve suggests talking about WSDL binding in place of banana calculus as people involved not present - Agreed
ACTION: Chairs to send Marco the requirements document (once approved) and ask if his requirements are met. (0301, 0309) DONE
Email sent to Marco: http://lists.w3.org/Archives/Public/public-ws-chor/2004Mar/0018.html
ACTION: Yves update the choreo web page to reflect the correct status of model overview as WD, which was done at last f2f in December i.e. it should be Published (0301, 0309) DONE
Yves, waiting for approval from co chairs of draft for submission dated March 15. Approved in call by Steve & Martin, so Yves will apply to get the update done to the site.
ACTION: editors to produce html of doc: http://lists.w3.org/Archives/Public/www-archive/2004Mar/att-0002/cdl_v1_editors.pdf (0301, 0309, 0316) IN PROGRESS
No progress
ACTION Yves to publish the html of WD of CDL (dependent on previous action. (0309, 0316) IN PROGRESS.
Awaiting completion of previous action.
ACTION: Yves add link to current editor's draft of CDL (will point to the WD for the moment). (0301, 0309, 0316) IN PROGRESS
Awaiting completion of previous action.
ACTION: SRT send email to public list reminding people how to handle comments and issues on the docs. (0309, 0316) IN PROGRESS
No progress.
ACTION: Kevin cast his comments in the form of issues and send to comments list (public-ws-chor-comments@w3.org ) (0301, 0309, 0316) IN PROGRESS
No progress
ACTION: Yves put reference to issues email address on public choreo page. (0301, 0309) DONE
On public page
ACTION: Yves will put a reference to the Bugzilla issue list in the public page DONE
On public page
ACTION: chairs schedule coord protocol issue (0301, 0309) DONE
On this agenda
ACTION: Greg to code up use case (from req doc) in CDL (0301, 0309, 0316) IN PROGRESS
Being worked on.
ACTION : Code banana calculus examples in CDL (0316) IN PROGRESS
No progress
Revisit of action items:
Far east meeting in sept/oct 2004. Group needs to decide if we do (or don't). Steve is concerned that progress is made. This is a function of who is present. Yves suggest use of WBS.
Martin suggests concentrations in Europe and US in order to maximize participants. If new members come from the far east, then we can revisit.
Steve R. Suggests idea of a ballot on where people would prefer to go. Yves suggests ballot with rating locations from 1 to 5 between east coast US, West Coast US, Europe and Far East.
Agree to run a ballot. If there is a tie, chairs will decide.
NEW ACTION: Chairs to set up ballot on preferred F2F locations.
Recap of meetings planned ... mid May 11-13th at Oracle, Redwood Shores; July options of either Loch Lomond or St Andrews (Scotland) - Steve will also look at London;
For July mtg, target is either wks beginning 19th or 26th July. 19th is a WSI mtg in Hawaii, so Martin proposing last week of July.
Jeffm:I have a f2f in boston thursday and friday of that week.
Jeffm: Also, you should know that week has also been prposed for wsd
Yves: how about a poll for availability at that time?
Mchapman: well early in the week will be impossible jet lag wise for those coming from hawaii
Jeffm: fly on sat, get there on sunday
Action item on issues list:
Martin. We have a bugzilla list that has been tidied up. Requirement items have been closed off which leave about 50 remaining. Need to do a triage ... are these well formed, are they still issues? So that we are left with a tidied up list.
Greg is in favour of having a sub group to do the tidying up. Agreed - need volunteers: Martin, Greg, Monica, David, Nick all volunteered ... but NOT STEVE !!!!!!!!
NEW ACTION: Martin to organize a call for the issues tidy-up sub-group.
Nick, Steve met with Robin Milner and ??? discussed the idea of an intermediate end-point language
Summary by nick: CDL has been a language that describes a global viewpoint between two or more participants
The idea is that you can project the model onto an individual participant. e.g. from the Buyers perspective, would be that the buyer sends a message to the supplier. The suppliers projection would be that they receive a message from the buyer.
Do we want just a global model, and leave the projection mechanism undefined, or do we want to define a mechanism for defining a projection and how it would be used.
Steve R. A projection is essential if you want to build tools.
David. The real debate is whether or not the method of doing a projection needs to be standardized (or not).
Tony. Can a projection be defined as an abstract BPEL process.
Steve R. That's a possibility but, what happens if you don't want to use BPEL.
David B. Do we or don't we need a standard end point language?
Nick: +1 on what Steve said
Monica. Would be able to generate an abstract BPEL process and is that within our scope?
Ugo. Problem of languages that do similar things. e.g. we have BPEL, Java, Ws Chor and others.
Yves: The tool can take WS Chor and project it to whatever is needed. Not sure of the benefit.
Martin. Not sure of the benefit, but if it is just a subset of the language then maybe it's a good idea.
Steve. Lot's of valid points. One of the goals is that you can generate stubs for Web Services in order to help guarantee compatibility.
Steve. Not sure what we actually have to do to do this.
Do we need an end point language or do we need to be able to demonstrate that we can produce such a language.
Steve R. What's the difference between developing a guide on how to, say generate BPEL, Java, etc. and defining an actual language
Nick. Not sure we need a separate language. The same language should be able to capture both common and end point perspectives.
Nick. CDL should be able to capture both. This was being discussed with Robin last week.
Nick. Pi calculus is an endpoint language. However you also need a higher level abstraction which provides a global perspective.
Nick. Says Robin is looking into the idea of having one language that can describe both.
David B. Suggests looking at some examples to work out if we can actually map to BPEL, Java, etc. Publishing the result is a separate decision.
Martin +1 for david
Nick. Discussion around end-point "contract" language. Abstract BPEL is an example. Java is not, or at least isn't adequate (Martin).
Martin, says that Java interface is a contract language.
Nick. Java interface is too little and full blown Java is too much
Ugo. Doesn't agree.
Martin. What we need is a solid proposal on what to do. Abstract ideas and opinions are not enough.
Steve. This is a hard idea to progress without a concrete proposal.
Suggested action: Steve/nick agree to take forward
Martin. Rather than action, it should be taken as an issue and link the proposal to the issue.
Martin. We have aired the issue and had some feedback. Its up to someone to make a proposal in order to continue the discussion in the future.
Steve. Issue is "It is not clear how to use CDL to generate an end-point language". Taken by Nick/Steve
Steve. This is important around state alignment - nick?
Nick. this is one tree ;)
Nick, Alignment is about making sure that both have a shared understanding of information.
Nick. The other, is that the processing of each participant is kept "in step". e.g. if one participant fails than the other participant(s) need to fail too.
Nick. The question is should the infrastructure take care of this, in terms of messages, or should the messages be described in the contract/CDL definition
Martin. Thinks its obvious that some type of coordination is needed. Question is do we standardize it in this group, do we allow other groups to do it or is it left as a vendor issue.
Tony. Don't think that we need to invent another protocol unless we can show that is no usable or protocol available to do exactly what we need to do.
Tony. thinks that examples given by nick can be handled by a transaction protocol.
Tony. Although we don't need to invent the protocol, we need to show in the CDL the points at which the protocol is started and terminates.
Martin. How will this help interoperability?
Tony. CDL by itself is not an interoperable protocol. CDL actually defines the flow, but not the actual messages.
Martin. This is just like any other service that needs to be defined.
Tony. WSDL does not give you interoperability, if complete, will provide the complete specification of the XML is sent.
Martin. This is only the concrete part.
Tony. When you actually use the language then you have to describe the message content. Question is how do you use the CDL for this purpose, or do we leave the hooks.
David B. question how do you bind CDL to multiple different coord protocols.
Tony. Should allow for pluggable coord protocols.
David. So the question is do we specify how to do one. IF we don't then the coord protocol will not be usable
Tony. Basically yes. At the high level it will be a "sketch map" description.
Martin. We've already agreed these levels.
Monica. Not sure if any of the coord protocols are mature enough. tony seems to be suggesting that we have an awareness of a coord protocol but that we don't have to map to one or define one yet.
Monica. we need to consider the opportunities of work done already.
David: +1.
Steve: +1 on pluggability
Steve. Let's summarize. Do we agree that: 1. We need plugability? 2. We are not choosing one over another?
David. + 1 on both
Monica. Perhaps this should be a "best practice" that we can cover in a whitepaper until specs become mature enough?
Martin. Whitepapers are not out of scope but are a separate decision once we decide what to do.
Steve. Endorsing Nick's request that we need some guidance ... suggests Tony ... plus others in the group.
Steve. Thinks a paper would be a good idea.
Tony. Offers to revamp paper in terms of new CDL (if he finds the time).
Tony. There's six coord specs out there. However all six have a common core with the same semantics. Differences tend to be in the detail and how certain corner cases are handled. So it should be possible to have an approach in the CDL that could map to all six of them easily.
Steve. Need a position paper that reviews protocols. Secondly, need a demarcation of start and stop. We should raise an issue against the CDL along the lines of "Coord protocol foobar can't be used with the current version of CDL".
Tony. Agrees to try.
NEW ACTION: Chairs to set up ballot on preferred F2F locations.
NEW ACTION: Martin to organize a call for the issues tidy-up sub-group.
Steve can't participate on April 20th.
<proud father> www.sixfoothighfilms.com/homecoming </proud father>