2003-20-FEB WSA Telcon -- 26 Feb 2003

Present: Daniel Austin, Mike Ballantye, David Booth, Doug Bunting, Tom Carroll, Ugo Corda, Roger Cutler, Paul Denning, Zulah Eckert, Gerald Edgar, Colleen Evans, Hugo Haas,Hao He, Yin-leng Husband, Mark Jones, Heather Kreger, Sandeep Kumar, Mike Mahan, Frank McCabe, Jeff Mischinsky, Don Mullen, Eric Newcomer, David Orchard, Katia Sycara,

Regrets: Mike Brumbelow, Chris Ferris, Suresh Damodaran, Martin Chapman, Geoff Arnold, Abbie Barbir, Duane Nickull

Chair: Mike Champion

Scribe: Chris Ferris

<scribe-cf> ACTION: Hugo to do Glossary - missing definitions from document in Glossary -- PENDING
... ACTION - Hugo to do add a priori to Glossary -- DONE
... ACTION - Hugo to do contact WSD document editor and resolve this issue -- DONE
... ACTION - Eric to incorporate these suggestions regarding the document draft -- DONE
... ACTION - Hugo to include the ebXML stuff that we propose to change, and post -- DONE
... ACTION: MikeC to explore subsetting issue with WSDL group. -- PENDING
... ACTION: Daniel to close 23 per his option 1, and open a new issue for "context" -- PENDING

<scribe-cf> MTF: final set of edits for submission to f2f; request time on agenda for f2f; will distribute draft by next friday
... MTF: requests 2 hours on agenda for f2f
... ACTION: Heather to send formal MTF submission to list by next week
... SOAP Features TF: some text written for document; making good progress
... ================
... Planning for the F2F
... make sure you have registered!
... http://www.w3.org/2003/01/TPMar2003/regByGroup#WSA
... 1 hour joint discussion with representatives of the P3P WG
... Presentation by interim editors on the refactored WSA document Discussions/votes/plans needed to get it out as a WD

<scribe-cf> hugo: deadline for registration feb 27
... mikec: if you have problems registering, contact chairs or staff contacts
... frankmcca: this is of the essence; anticipates lots of
... discussion; propose presentation and then have discussion
... mikec: 1 hour preso followed by couple hours discussion?
... katia: 3 hours, yes
... dbooth: break that time up?
... frankmcca: have feedback session towards the end
... mikec: focus on getting material into the document
... mikec: where do we stand on the issues list?
... TomCarrol: hugo has closed 2 and there may be a 3rd pending
... daniel sent some proposals...
... mikec: ready to close all but one
... TomCarrol: another issue; describing web services stack and coordinating that against OSI stack
... mikec: an hour at the f2f to close any we can close?
... TomCarrol: haven't seen status; an hour should be fine
... mikec: anyone else need time at the f2f?
... any other major things? focus is on document; some coordination issues; any glossary type things
... mikec: if we had another 1/2 day breakout; any other subjects?
... colleen: do we need pftf to give a report?
... mikec: how much time will be needed?
... scott: think don's plans to be at plenary, but not attending arch WG
... colleen: let's put placeholder, I'll follow up with Don
... Roger: relationship to ebXML?
... mikec: there's been enough traffic; reasonable breakout topic
... martinc: can we do this with the stack discussion?
... chrisf: thinks its a different discussion
... mikec: thoughts on time on agenda for choreography issues?
... martinc: can't say much because the wscwg f2f is the following week
... mikec: then 10 minutes to give us your thoughts
... martinc: that's fine
... mikec: any thoughts about web arch vs web services arch?
... frankmcca: that'll be one of the discussions relative to the refactored document
... katia: maybe at lunch a smaller set of people; there's a harvesting from REST group that has to report
... abbie: what about security?
... mikec: yes
... abbie: yes, we need a section on that
... mikec: could you volunteer to lead a discussion?
... abbie: I could do that
... we have to look at interface with identity; these will be consumers of web services
... mikec: don't have to prepare formal discussion; maybe point us to resources and we could have hour at f2f
... abbie: i can do that
... mikec: anything else?

<scribe-cf> hugo: more directions we should be heading; we missed a heartbeat on the 14th; would like the f2f to be focused on getting something published soon
... we should also have time to plan and/or approve schedule for publishing
... chrisf: would be willing to take a stab at putting the html into xml-spec dtd
... mikec: katia and frank; you planning on continuing in this process?
... frankmcca: depends on what happens at f2f
... next few weeks don't have huge number of cycles
... katia: would be willing to continue working; there are some parts not in sect 2 that needs refactoring
... I would be willing to continue working
... mikec: would working in XML be a productivity hit?
... chrisf: we have free license to XMLSpy; I can dig up URL of email from altova
... Roger: we have XMetal too, right?
... frankmcca: is there a MacOS version?
... Roger: no
... frankmcca: is it in Java?
... Roger: no
... mikec: once we get into more of a mecanical mode, would be more productive to do work in xml-spec dtd
... mikec: eric still hasn't joined us; frank or katia, could you lead us through this discussion?
... in 2 weeks, we want to have serious discussion of this
... frankmcca
... sent a diagram earlier today
... maybe we can walk through it?

<scribe-cf> frankmcca: not intended to be complete; also slightly inaccurate
... mikec: concepts are organized aplhabetically
... frankmcca: some people see things pictorially, others textually
... dbooth: high level view of things
... likes the idea
... frankmcca: like to go through it; center of diagram is service; service forms a task; involves executing goals
... agent requests and provides service
... dropped notion of actor
... service has description; describes semantoics and interface; doesn't need to be formal
... services exchange messages; message has sender and receiver; messages trigger or result in actions which is how the link between mesages and tasks are established
... links into manage,ment
... deployed element also owned by legal entity
... also associated with management interface and has lifecycle
... to do the whole thing complete justice, have number of diagrams each a different aspect

<scribe-cf> Roger: jjdubray sent out a UML diagram of this
... frankmcca: we should discuss this diagram; his is also useful
... purpose of diagram is visual reference for architecture
... frankmcca: shows relatoonship of what is going on
... not intended to be definitive diagram
... katia: diagram may also help us to identify what are important concepts and we can dsrill down in subsequent diagrams in the sense of adding relationships that are not as important
... dorchard: really likes this diagram
... thinks it provides lots of clarity of relationship to terms because we can get into specific discussions about what we should or should not talk about
... goals and actions we should not talk about becasue we cant write specs about them, etc.
... suggest that the things that are in three boxes in right hand corner could be removed
... can talk about scope
... mikec: remove altogether?
... dorchard: can talk about things from bits on wire; goals are much more ephemeral
... everyone says high to Eric
... Roger: agree bottom three may be out of scope; one thing that leaps out at me is why is deployed agent at thesame level?
... mikec: do authors of this diagram want to respond to david?
... frankmcca: without talking about some sort of action, you don't have service
... we cannot and should not talk about internals of service
... inherent to semantics of service that things happen as a result of using the service
... no need to go into huge amounts of detail
... katia: lots of discussion about representation. action is ...
... dorcharda: how to do this discussion
... there is a queue
... dbooth: go ahead dave
... dorchard: one pushback from issue raiser
... guess point that invocation of service results in some action, but we cannot say what it is
... katia brought up issue of what action means from visibiulity perspective, serivce is distinct from message
... different concepts
... katia: agree; if you look at document in terms of concepts
... Roger: more I look at this, more I think it is wrong
... mikec: what is next step on resolving lower right hand corner
... Roger: looks like it is just wrong
... deployed element is paralell to agent
... frankmcca: this is coming from input from mtf
... the management folk want to be able to manage services and agents, they also need to manage descriptions
... mikec: roger, you want to initiate email discussion
... katia: deployed element actually a resource, lets call it element... in heat of getting this out, may need discussion
... mikec: lets make sure these are separate threads
... ACTION: Roger to start email discusion
... dbooth: there will be questions abou the technical correctness of some parts, inclusion of lower right hand corner, personally think we shouldn't worry about that
... if some things are out of scope, we can grey them out in diagram
... visually indicates what is in and out of scope
... dougb: not entirely certain why we want to spend that part of our diagram that which is out of scope
... mikec: it is here but we aren't going to talk ablout it
... dbooth: easier to understand scope if you say what is in AND out of scope
... ACTION: DaveO to spawn email thread on lower righthand corner
... katia: that can be assumption that is made explicit, otherwise one has lots of discussion about how to
... sandeep: two comments, provisioning a service dores this mean it is managable component? my understanding lower righthand corner describes semantics of service
... if dotted link from semantics
... frankmcca: yes, should be link
... sandeep: two diffeerent aspects
... katia: think that is great
... dorchard: one of things I saw was that message not directly related to description, but was related to action
... wsdl describes message structure
... wsdl describes the interface, may describe semantics, may describe locations... think service location may be missing
... mikec: interface should point to message?
... buncha yesses
... dorchard: wsdl also describes service endpoint or location
... katia: endpoint is binding of port to a location, need notion of port
... this needs to be put in
... not there yet

<zulah> Sandeep's question about provisioning a service - the term "manageable" implies that the element has a sufficient set of capabilities such that it can be managed. In this context, provisioning will be enabled for a manageable element.

<scribe-cf> ports and endpoints as managable elements
... dorchard: don't think you need to copy wsdl domain model
... theres a service interface, semantics and location
... don't think we need to talk aboyut it in our model
... katia: does not capture all relationships in our docuemnt
... dorchardL it is in our web services definition
... mikec: service has identity/identifier
... katia: different relationships
... mikec: do editors accept these points?
... anything need to be taken to email?
... katia: discussion of action
... frankmcca: not clear on something; the interface to me is the abstraction that denotes how you interact with the service
... eric: just serice name and data
... frankmcca: relationship between exchange and interface
... eric: not sure a service has an MEO
... MEP
... katia: a service knows which messages it dores not understand
... frankmcca: think I have it
... service has identifier
... pattern of messages and location of service
... mikec: can you take this under advisement and send out new diagram?
... dbooth: don't see client
... katia: requestor?
... eric: agent that requests service
... mikec: consoistent with way people use term agent?
... mikec: essentially two aspects of same thing?
... eric: service can invoke another
... dbooth: ooh, ouch
... eric: client is separate thing?
... dbooth: associate service with client/service relationship
... eric service can be client of another service
... dbooth: would use term agent for that
... frankmcca: focus is on the serivce providers POV
... maybe need complementart POV?
... mikec: like franks suggestion
... katia: like your suggestion
... this is like ERD
... this is not the only diagram; need to create additional diagrams for different POVs
... dbooth: may have nmisunderstood "intent"
... eric: slippery slope if client is something different
... dbooth: agree with concept of something that encompasses both
... eric: then we change word for service, it is the same thing
... dbooth: generically would call it an agent
... like left and right
... eric: can something have one role? would say answer is no
... frankmcca: don't need to constrain
... eric: don't see how a service can't be client or vice versa
... frankmcca: service is about potential for action
... agent can provide and request services
... eric: bad definition
... any agent should be capable of any behaviour
... sooner or later, there will be callbacks, etc.
... katia: look at actual definition, all we are saying a service may be a provider, requestor,,, not precluding either
... it is agnostic
... dorchard: think that I understand dbooths pushback, also with eric
... agent can play multiple roles, think thats a good way of showing it
... in web arch, there is server and user agent
... need a term that encompasses both, we used agent
... think this accurately reflects this
... like to keep that off this diagram
... there was some talk about whether an agent was a realization of a service
... a service description provides a URL a description describes a service, the description has a URL
... it is the thing
... I couple the instance in term service as well
... eric: you distinguish between service and descriptions as separate entity
... dorchard: we combined them and called it a web service
... katia: as general point, wsdl, soap and uddi are example technologies we use today
... they are not the objective definition
... we should not say this definiton is wrong because wsdl defines it this way

<scribe-cf> dorchard: if we end up saying in our domain model that there is a separation of terms and then have this problem of figuring out how that maps onto wsdl, we havge to think about whether we want to go there
... mikec: we're out of time
... dbooth: believe daveO liked a concept that could encompass either role. I like that also
... I understood use of word service on here
... service is too strongly associated with one end
... dorchard: what is the thing that sends a solicit response message
... dbooth: absolutely
... they are relative terms
... you could view it either way
... left and right are relative terms
... don't think it should be that term; should be agent; in diagram frank distinguishes between agent and service
... adjourned
... we be jammin'

<chrisf> ACTION: Mike to do recruit members to work on Usage Scenarios Document as co-editor -- PENDING
... ACTION: DaveO to send some notes re REST RPC scenarios. -- PENDING
... ACTION: wsa-members to do all note-takers from break outs to send notes to w3c-ws-arch list -- PENDING

2003-20-FEB WSA Telcon

Attendees

Contents

Summary of Action Items