Present:
Regrets:
David Booth W3CChair: Jonathan Marsh
Scribe: Roberto Chinnici
Scribe: Minutes from last week are approved.
Scribe: PENDING ACTION: 2003-01-21: Roberto and gudge to
create a branch and work up a binding proposal based on referencing
type systems directly from operation components. (Umit's example,
Sanjiva's example, WSDL 1.1 example, and others.)
... DONE ACTION: 2003-02-27: Sanjiva to send summary of
one-portType-per-service issue.
... PENDING ACTION: 2003-03-04: Editors to discuss markup for testable
assertions in the spec and come back with a strategy.
... PENDING ACTION: 2003-03-04: Jonathan to recruit a QA contact for
the WG.
... PENDING ACTION: 2003-03-04: Jonathan to recruit a test contact for
the WG.
... PENDING ACTION: 2003-03-13: Editors will find part 2 issues to
dispatch easily next telcon.
... PENDING ACTION: 2003-03-13: Don will write a proposal for
annotating schema with part information.
... DROPPED ACTION: 2003-03-27: Jonathan will follow-up with editors to
figure out how to improve the prose of the spec to be aligned with
schema.
... PENDING ACTION: 2003-03-27: Philippe write up a proposal for
embedding binary data types in schema
... DONE ACTION: 2003-04-03: Arthur to bring discussion to group in two
weeks (more or less) for solutions to R085
... DROPPED ACTION: 2003-04-03: Arthur to coordinate work on WSDL
validator
... DONE ACTION: 2003-04-03: Editors to include normative schema
language in spec (conformance section?); schema to be separate, in TR
space.
... DONE ACTION: 2003-04-03: Jonathan to respond to OWL with "no plans
to review, no resources, little knowledge of why requested".
... PENDING ACTION: 2003-04-10: Sanjiva to rewrite his proposal on
bindings.
... PENDING ACTION: 2003-04-17: Gudge dig out MIME type related parts
of Proposed Addendum to SOAP with Attachments proposal (as sent to
XMLP) and post to WSDesc Discussion list
... DONE ACTION: 2003-04-17: JMarsh to respond to XML Schema's request
for review
... DONE ACTION: 2003-04-17: Jacek to make a proposal for issue 28
... PENDING ACTION: 2003-04-17: Sanjiva to take the lead on
coordinating the meeting on bindings the day before the F2F
... DROPPED ACTION: 2003-04-17: Sanjiva, Philippe and Arthur to meet
the day before the F2F to talk about bindings
... DONE ACTION: 2003-04-17: Youenn to see about getting a speaker
phone for the F2F
Youenn: the meeting will be at the hotel; internet access in
the meeting room.
... train station - hotel distance is 500m/1000m.
JM: any concerns?
Arthur: only impact currently on hospitals and hospital workers.
JM: should monitor the situation in the next few weeks.
... moreover, WHO's decision was controversial.
JM: looking for volunteers on the West Coast
JM: providing an ontology would give us RDF for free.
... we don't have anybody appointed to get the RDF mapping done.
... DBooth potential candidate; Gudge has RDF background too.
Gudge: not interested in getting involved.
Jacek: trying to be involved in RDF but can't serve as editor.
Scribe: ACTION: Jacek to look at OWL and report back to the working group.
JM: will try to get somebody from OWL to talk to this working group.
plh: we could invite an expert to the FTF meeting
Arthur: we don't have to do an OWL mapping
JM: it might turn out to be more convenient, for a number of reasons.
Scribe: ACTION: JM to get somebody from OWL to talk to the working group.
JM: goal of the FTF to get ready for the next publication.
JM: any objections to declare QoS as out-of-scope?
Scribe: No objections.
Scribe: http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0052.html
<alewis> http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0080.html
Scribe: Discussion to try and clarify the issue.
Arthur: "semantically equivalent" endpoints -- what does that really mean?
JM: seems tied to Sanjiva's proposal.
Amy: it is a new issue though.
Resolution: Add new issue on multiple endpoints with the same interface
Scribe: Jacek's proposal at http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0064.html
Jacek: proposal is to rename the "transport" attribute to
"protocol".
... change the value to the protocol binding name in the SOAP 1.2 spec.
Resolution: proposal is accepted, issue 28 is closed.
Scribe: http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0069.html
Sanjiva: allow only port type per service.
JM: are Steve Tuecke / Steve Graham OK with this?
Scribe: They are concerned about the same issues that Amy raised.
Amy: there is no way currently to indicate that two services
are connected in WSDL.
... could live with either solution.
Umit: Sanjiva's solution seems to require references within WSDL.
Arthur: passing references to other services would help.
DonM: it wouldn't work with all bindings.
Arthur: operations would return a reference typed to a
portType.
... implementors could choose whatever binding they want.
Gudge: service with two portTypes with operations returning an enumerator and a set of items: would I need to write two services?
Arthur: no, you can have a single interface inherit from the two ones you described.
Gudge: it doesn't seem to simplify things really, because you have to define a new interface.
Arthur: no classes in WSDL, only interfaces.
... analogy: service is an instance of a class, so it needs a type.
Gudge: what about interfaces? you can implement multiple ones without bringing them together in a single one.
JeffM: a discussion on multiple interfaces is deja vu. it
centers on the notion of identity for services.
... is it two implementations of the same thing?
Amy: issue is: should we force people to define an aggregate interface?
Gudge: by inspecting a service, you can find a list of
interfaces it implements.
... if all those interfaces had a single ultimate base, you could
discover that.
Sanjiva: bindings must completely implement an interface.
Gudge: people who want a single interface per service can already do that.
Amy: today we have "interface sets" for services and we shouldn't restrict that.
JeffM: how can I tell if I have references to different endpoints of the same service? it requires a notion of service identity.
Amy: Sanjiva's proposal doesn't address JeffM's issue, although Arthur's does (up to a point).
Roberto: The simplification is a good idea. Restricts the
notion of what a service is and this rationalises things.
... Proposal provides a single view of service and will make
relationships between services explicit
<jeffm> +1
<sanjiva> +1
... I can make a proposal for relationships (these are what UML
considers associations). Could try to get it ready for F2F.
JeffM: in favor of the simplification, but we need to address
the identity issues.
... writeup refers to WS-Addressing; can we even talk about it here?
IPR?
... discussion is really about a component model.
Arthur: we could define a standard interface with a
getIdentity operation.
... implementing this interface would make your service "identifiable".
Gudge: what would the component model look like?
Sanjiva: (describes change in proposal)
Gudge: it doesn't seem sufficient -- there is more (service->bindings->portType)
Sanjiva: it can be made to work
Gudge: but there are other changes
<jeffm> IMHO - I could be completely confused - but I think we need to nail down the concept and semantics of a service-ref (type or instance for eg). It is fixable, but it has to be done before we're done
Arthur: port needs to implement the interface declared in the
service element, but it could have additional operations (or not,
that's a decision point for us)
... you couldn't implement less than the interface declared in the
service.
JM: strawpoll
... do more work on Sanjiva's/Arthur's proposal?
Result: 10 for, 8 against
Scribe: Detailed results are as follows
... Yes (10) - Erik Ackerman, Allen Brookes, Roberto Chinnici, Steve
Graham, Jacek Kopecky, Philippe Le Hegaret, Jeff Mischkinsky, Arthur
Ryman, Sanjiva Weerawarana, Umit Yalcinalp
... No (8) - Dietmar Gaertner, Martin Gudgin, Amelia Lewis, Lily Liu,
Dale Moberg, Don Mullen, Jeffrey Schlimmer, Jerry Thrasher
SteveG: service identity problem is hard, lots of work on it in OGSA
JeffM: it's a problem for all distributed systems, let's look at prior solutions.
Arthur: look at existing endpoint reference issue.
... it's R85.
JeffM: can we talk about WS-Addressing?
... IP on WS-Addressing is cloudy. IP is not copyright.
JM: we can still talk about it.
JeffM: worried about discussing a proprietary proposal in this context.
Arthur: requirements in WSDL have been out (public) for a long time now.
JM: no requirement to deal with WS-Addressing in particular; R85 is fairly general.
<Arthur> could Philippe give us the official W3C guidance on discussing WS-Addressing?
JeffSch: there is a process to contribute a spec to the
working group.
... there is also a different mode where people with knowledge of an
external spec bring the subject up for discussion in a working group.
Arthur: treat WS-Addressing just like Grid.
JeffM: Grid has a constituency of a different kind than a private group of companies.
JeffSch: people bring ideas to the wg for discussion all the time.
JeffM: random ideas and specs with unclear IP are very different.
<Arthur> the proposal is not to adopt WS-Addressing, but to enable it just as we enable, say, XLink
JM: is it a concern that by reading the spec we could get polluted by the IPR?
JeffM: yes.
JM: the IPR exists even if we don't look at the spec.
... please read Arthur's proposal and we can discuss any eventual IPR
problems there.
Arthur: can we get a W3C statement on it?
JM: question is: can W3C members use specs with IPR attached to inform a discussion?
plh: no comment
JM: meeting is over