See also: IRC log
Present: bijan, mario, Dbooth, Mike_Champion, Hugo, Roger_Cutler, PaulD, Suresh, Gerald, Frank, S_Kumar, Katia_Sycara, Shishir_Garg, Abbie, Hao
Scribe: * Topic: Approval of Minutes
... Minutes of Dec 18 approved.
<mchampion> ACTION: Abbie will send suggestions to Frank and public list for incorporation of policy into other models. [PENDING]
<mchampion> ACTION: David to incorporate Management text from Zulah's last F2F presentation into document. See http://lists.w3.org/Archives/Member/w3c-ws-arch/2003Dec/0036.html
Scribe: ACTION: DBooth to address concerns about Management model redundancy with other parts of model in editorial note. [PENDING]
<Scribe> ACTION: DBooth to include text should put some precision around management interface. [PENDING]
<Scribe> ACTION: Frank will straighten out resource/service/description/publish, etc by tomorrow, [DONE]
<Scribe> ACTION: Frank to send out new text for Service model
<Scribe> ACTION: Yin Leng will look at EDI tracking text and see if something about this should be put into the Management text. [DONE]
<Scribe> ACTION: dbooth to send a message to the WSD WG asking if two WSDL documents can reference the same service [DONE]
<Scribe> ACTION: [PENDING] Bijan to check if someone from U.Maryland has the resources to help with OWL related work [DONE]
Frank: Policy model has been the most stable. Message model is close to stable. Service model is not yet stable. Don't bother looking at that yet. You could look at the Resource model though.
... Best to focus on the Policy and Message models.
Scribe: ACTION: Katia to review Policy and Message models and update OWL by next week.
Katia: Do we need to update the OWL after the WG ends?
PaulD: We should add it as an issue.
Scribe: ACTION: Katia, Bijan and Frank to figure out a process for continuing to update the OWL through the life of the WG and potentially after
<Scribe> ACTION: Frank will discuss with others how to refactor SOM [DONE]
<Scribe> ACTION: Katia to propose clarification of action/task/goal. [PENDING]
<Scribe> ACTION: Katia to send comments on Discovery text to dbooth and list [DONE]
<Scribe> ACTION: Katia to submit text on peer-to-peer discovery
<Scribe> ACTION: [PENDING] Mike to add and wordsmith text in 3.11 choreography
<Scribe> ACTION: [PENDING] Mike to propose changes to WS Reliability section in stakeholders perspective
<Scribe> ACTION: [PENDING] MikeC to ping DavidOrchard for his input on the Resource
<Scribe> ACTION: [PENDING] MikeC to propose text to replace these next three paragraphs and the diagram about SOA and distributed objects
<Scribe> ACTION: PaulD to propose text on federation of registries [DONE]
<Scribe> ACTION: Suresh to review doc and propose text on EBXML [PENDING]
<Scribe> ACTION: dbooth to clarify term "service provider" and "service requester" and expand glossary [PENDING]
Suresh: Re: Concepts and relationship "spaghetti" diagrams, I didn't see explanation of what the boxes and arrows mean.
Frank: It's at the beginning of section 2.
Suresh: It doesn't define the meaning of the arrow though.
Scribe: ACTION: dbooth to explain the meaning of an arrow in mind maps
Suresh: There's a discussion of messages, and one of services, but not a connection between them.
... This is in the "Overview of Engaging a Web Service".
Scribe: ACTION: dbooth to look at better explaining the connection between messages and services in "Overview of Engaging a Web Service"
Frank: One of the objectives of the revised service model was to make that clearer.
Suresh: In the abstract, typo in last sentence (missing "component").
... Also look for the word "standard" and remove all occurrences.
... I was amused by the idea of an agent. Sec 1.5 "What is a WS" says "machine-to-machine interaction". I think "between agents" would have been better.
Scribe: (Several people noted that this def was hard enough to reach, we're not apt to change it.)
Suresh: Incongruent definitions: sec 184.108.40.206 says an address is info used to identify how and where to deliver messages. Might want to say "when a specific mechanism is used".
... Sec 220.127.116.11 on message correlation: I struggled with it. Also the relationship between the correlation and MEPs. That paragraph could be improved.
Roger: That was another paragraph that resulted from intense negotiation.
... You end up with something that nobody thinks is idea, but people can swallow.
... Sec 2.3.2 on service task: Is that a post condition?
Katia: No. I have an action to work on that.
Suresh: The idea of agent was interesting. And I liked the pictures. Overall I thought it was quite well written, but lots of incongruence.
MikeC: Hardest thing will be to change the concepts and relationship, because they need to be synchronized between the text, diagrams and OWL.
... Send any proposed changes NOW.
Scribe: By a few days before the F2F, the document *will* be in its final state,
... net wordsmithing, editors notes, and
... we'll have to delete anything we can't agree on.
... At the F2F we'll step through the document in a time-boxed way.
Roger: Why not say "this is our last call"?
MikeC: That's essentially what we're doing, but can't call it W3C LC because there's a list of criteria that we don't meet.
Katia: I agree.
Hugo: The idea of LC is that you have addressed all issues, met all requirements, and met all dependencies. You've addressed all the charter asked you to do.
Roger: I think we have.
Hugo: I think we have a set of substantial issues that we cannot easily address.
... Also, I don't think you can put a spec to LC and run away. You need to be able to address comments, etc.
Roger: I don't see why we can't see all those things.
Scribe: Process doc on LC: http://www.w3.org/2003/06/Process-20030618/tr.html#last-call
MikeC: I would entertain a proposal.
Scribe: ACTION: Hugo to research feasibility of advancing to Last Call
Hugo: One thing that needs to be done is to go through every requirement and verify that it has been done.
... Also need to go through every issue and respond to the person who raised it.
Roger: Sending somethign to those people is better than ending the WG without saying anything.
MikeC: People who care about doing this should volunteer to do it. Vote with their feet.
Katia: To what extent does the goal of something like like security (for example) need to be addressed?
Roger: There we a lot of AC reps that said WS Arch is important to do, and some that were dismayed at the WG's closing. We should at least give it a chance.
Scribe: ACTION: Katia to take inventory of requirements to see what we've and what we haven't met
<Scribe> ACTION: PaulD to take inventory of the issues list to see what we've met or not met
Roger: Would be nice if WSD takes over the use cases, but since they're broader than WSDL concerns, and Hao did a lot of work, I think we should publish it as a Note.
MikeC: I think we should publish it.
Roger: I think he's working in xml-spec.
Scribe: (Group agrees)
... [Hao joined the call at this point.]
<Scribe> ACTION: Hao to produce a version of Use Cases suitable for printing prior to the F2F
<Scribe> ACTION: MikeC to put Use Cases on the agenda for next week
Frank: Looks perfect to me. :)
Katia: Looks ok to me.
Frank: BTW, I changed "legal entity" to "person or organization"
... Should we incorporate the trust diagram in with this?
... It took a while to integrate policies into the rest of the arch. It may also take long for trust.
... Highest priority at present is the service model.
Frank: Current version has the new diagram, though text is not yet correct.
... We need parity on task/goal/action/(state).
... Also, on intermediary, our story isn't sufficient. The diagram I have talks about service roles, which are associated with tasks.
... Other than that I'm much happier with this model of service than the previous one.
... Messages have aspects that are relevant to the service role.
dbooth: that sounds slippery to get into.
Frank: There may be more than one service that looks at the same message. Therefore, what is the relationship of service to message?
Frank: Think of a document that is being passed around. You can imagine a whole bunch of things that happen to it, but the document has a life that is indepent of the things that act on it.
MikeC: If you write a check, you shouldn't care if it's processed by hand or machine.
Frank: And it's still the same check, whether it's the physical check or a scanned image of it.
Roger: I'm confused when you say a message is the same message if you say it is.
MikeC: Let's critique the text when we see it.
Frank: think of an encryption intermediary. From the POV two endpoint principals, the message is the same. But from the POV of others, it is different when it's encrypted.
Katia: But that would be a translation of the message.
Frank: Another example: If you have a document representing an expense claim and it needs to be processed by a number of people.
... It's in electronic form, but needs to be passed along through a chain of approval.
Roger: Suppose A sends message M to I. I sends it to company C who say's it will fill the order. I sends the same message to D, but D doesn't recognize it.
Frank: The point is that people agree that it's the same message.
Katia: I was confused about "aspect". When one thinks of the service model, one doesn't think of intermediaries. One thinks "I want to buy a tank".
... We can then say intermediaries complicate the picture.
Frank: The fundamental concept of "intermediary" is that the message it sends is in some sense the same as the message it received.
Katia: We should make the diagram more explicit about including "intermediary".
... The picture needs to be self-explanatory in some sense.
MikeC: The other alternative that the concept of intermediaries is a wart that we should admit.
Katia: The current diagram is not capturing intermediary.
dbooth: I think we could do without "aspect" in the service model.
Frank: But then how do you account for more than one service processing a message?
dbooth: I think it's an issue of how a WS is used -- not architectural.
Scribe: (No objections to DBooth's proposal at http://lists.w3.org/Archives/Public/www-ws-arch/2004Jan/0002.html )
... [Meeting adjourned]