- From: Jonathan Marsh <jmarsh@microsoft.com>
- Date: Wed, 7 May 2003 11:40:41 -0700
- To: <www-ws-desc@w3.org>
[Revised, original minutes at http://lists.w3.org/Archives/Public/www-ws-desc/2003May/0005.html. Corrections at: http://lists.w3.org/Archives/Public/www-ws-desc/2003May/0010.html, http://lists.w3.org/Archives/Public/www-ws-desc/2003May/0011.html. See also http://lists.w3.org/Archives/Public/www-ws-desc/2003May/0013.html.]
Minutes, 1 May 2003 WS Description WG telcon
Present:
Erik Ackerman Lexmark
Mike Ballantyne Electronic Data Systems
Allen Brookes Rogue Wave Software
Roberto Chinnici Sun Microsystems
Glen Daniels Macromedia
Steve Graham Global Grid Forum
Martin Gudgin Microsoft
Tom Jordahl Macromedia
Philippe Le Hégaret W3C
Amelia Lewis TIBCO
Lily Liu webMethods
Jonathan Marsh Chair (Microsoft)
Jeff Mischkinsky Oracle
Dale Moberg Cyclone Commerce
Arthur Ryman IBM
Adi Sakala IONA Technologies
Jeffrey Schlimmer Microsoft
Jerry Thrasher Lexmark
William Vambenepe Hewlett-Packard
Sanjiva Weerawarana IBM
Umit Yalcinalp Oracle
Prasad Yendluri webMethods, Inc.
Regrets:
David Booth W3C
Youenn Fablet Canon
Dietmar Gaertner Software AG
Jean-Jacques Moreau Canon
Don Mullen Tibco
Igor Sedukhin Computer Associates
--------------------------------------------------------------------
Agenda
1. Assign scribe: Jeffrey Schlimmer
--------------------------------------------------------------------
2. Approval of minutes of April 24 telcon [.1].
[.1] http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/att-0097/minutes_0424.html
Approved
--------------------------------------------------------------------
3. Review of Action items [.1].
PENDING 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.)
PENDING 2003-03-04: Editors to discuss markup for testable assertions
in the spec and come back with a strategy.
PENDING 2003-03-04: Jonathan to recruit a QA contact for the WG.
PENDING 2003-03-04: Jonathan to recruit a test contact for the WG.
PENDING 2003-03-13: Editors will find part 2 issues to dispatch
easily next telcon.
PENDING 2003-03-13: Don will write a proposal for annotating schema
with part information.
PENDING 2003-03-27: Philippe write up a proposal for embedding binary
data types in schema
PENDING 2003-04-10: Sanjiva to rewrite his proposal on bindings.
DONE [.2] 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
PENDING 2003-04-17: Sanjiva to take the lead on coordinating the
meeting on bindings the day before the F2F
DONE 2003-04-24: Jonathan to get somebody from OWL to talk to
the working group.
... Speaker is from University of Maryland and may join the WG to
... contribute to RDF mapping
PENDING 2003-04-24: Jacek to look at OWL and report back to the
working group.
[.1] http://www.w3.org/2002/ws/desc/#actions
[.2] http://lists.w3.org/Archives/Public/www-ws-desc/2003May/0002.html
------------------------------------------------------------------
4. Administrivia
a. May FTF [.1] - look for first draft agenda this week.
Probable joint session 1/2 day.
Do we want to go till 5 on Wed?
[.1] http://www.w3.org/2003/ws/desc/3/05/f2fMayLogistics.htm
... agenda pending
... joint session with arch on Wednesday PM, probably adjourn by 3:30.
b. July FTF. WHO has cancelled SARS warning - no plans to relocate.
... SARS warning cancelled for Toronto
... no plans to relocate
<plh-home> http://www.w3.org/2002/ws/desc/3/05/f2fJulyLogistics.htm
... Wednesday to Friday
... start time: tentatively start Wednesday morning
... joint sessions TBD
... W3C\Philippe updated page(s) with correct dates / times
c. Sept FTF [.2]
[.2] http://lists.w3.org/Archives/Member/w3c-ws-desc/2003Apr/0049.html
... holiday starting sundown on Friday
... tentatively Monday - Wednesday, 22-24 September
d. TF status
... no update from MEP task force.
... update from IBM\Sanjiva on context proposal in F&P task force.
... issue / question list and merge pending.
... 2 more telecons before F2F.
... expect to review during F2F.
... may result in best practices.
... interested to see relationship w/ OWL and/or RDF.
e. OWL last call review [.3]
[.3] http://www.w3.org/TR/owl-features/
... pending Systinet\Jacek
f. Schema subset feedback [.4]
[.4] http://lists.w3.org/Archives/Member/w3c-ws-cg/2003Apr/0010.html
... would WG be a potential customer for a Schema subset?
... IBM\Arthur notes JAX-RPC calls out a subset that isn't interop
... IBM\Arthur notes that WS-I does not identify a subset
... Oracle\Umit agrees
... Philippe notes that if there are interop problems, would prefer to address them soon.
ACTION: Sun\Roberto to write up uniqueness problem in Schema
... IBM\Arthur interested in exploring rationale behind required/optional motivation in JAX-RPC
<Roberto> Schema allows global validity constraints, such as uniqueness and keyref, and those are fine in full documents. But in SOAP we deal with XML fragments that are embedded in an envelope, so it could happen that fragments that are valid when considered as stand-alone documents become invalid when they are placed side-by-side in a SOAP envelope. Notice also that if those fragments went into attachments, they'd revert to being valid. Of course, the opposite
... Macromedia\Tom asks why WS Description WG needs to be involved?
... Microsoft\Gudge notes that lacking a WS-I profile for Schema in WSDL 1.1, wouldn't a profile over Schema for WSDL 1.2 be an unexpected restriction?
... Oracle\Umit sees a need for a subset, even if it has not yet been defined.
... Oracle\Jeff and Microsoft\Gudge discussed the degree to which the WS-I BP 1.0 explicitly did not take up the subset activity.
... Oracle\Umit notes that some aspects of Schema are not well understood and thus not implemented.
... IBM\Arthur suggests a good subset could be referenced by JAX-RPC and/or defined by WS-I.
... Chair\Marsh suggests we will point XML Schema WG at JAX-RPC and include Sun\Roberto comments.
<TomJordah> Tomj says as an implementor we would love a subset, but as a working group member I am unclear as to how we would make use of a subset
ACTION: Marsh to summarize WS Desc interest for XML Schema subset to XML Schema WG
------------------------------------------------------------------
5. New Issues. Merged issues list [.1].
- Kevin: "recursive" interface inheritance [.2]
- Arthur: R085, Relative URIs, and XML Base [.3]
[.1] http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/issues/wsd-issues.html
[.2] http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0155.html
[.3] http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0142.html
... recursive interface inheritance
... consensus that it is not allowed
ACTION: Editors to clarify that recursive interface inheritance is disallowed.
... IBM\Arthur post on xml:base
... IBM\Arthur asks whether we get xml:base for free, or do we need to explicitly support it?
... Chair\Marsh and W3C\Philippe pointed out that xml:base AII must be allowed by Schema -- explicitly or via wild card
... Microsoft\Gudge noted XMLP WG has an attribute wild card on soap:Envelope, Header, and Body
... and thus allow xml:base AII
... expect to discuss this issue in e-mail.
------------------------------------------------------------------
5a. Obsolete issues. The editors suggest the following issues are obsolete and request the WG blessing to close them:
7 Example incorrectly uses xsd:binary
9 Example misses "Soap"
10 Example 3 element order bug
11 Bug in grammar for <import>
50 SOAP examples declare arrays using an old schema definition
70 Error in HTTP GET/POST example
In addition, the following issues have been closed per the spec or ed todo list, but not closed in the issues document:
issue allow nonxml typesystems Allow non-XML type systems?
24 Real difference between literal vs. encoded?
25 Unclear relationship between XML Schemas and SOAP data model
13 Parameter Order missing from schema
22 Specification not XML Infoset based
16 Does a binding have to specify all the operations in a portType?
20 Inconsistency in definition of soap:header' (contains 'part' or 'parts' attribute?)
49 Inconsistency in "soap:header" specification
51 Asymmetry between soap:body and soap:header
... no immediate concerns on closed issues from WG
------------------------------------------------------------------
6. Proposal for restricting a service to a single interface [.1]. Some parts of the question:
- What is the motivation for the proposal? (Not clear from the
archive)
- Should we make clear that a <service> represents a single
entity (i.e. ultimate destination node in SOAP parlance)?
- How do we indicate that ports within a service are alternatives?
Don't we already have this information through
port -> binding -> interface?
[.1] http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/0069.html
... IBM\Sanjiva claimed that ==1 interface per service is simplest
... Tibco\Amy pushed back that it does not simplify overall WSDL document
... IBM\Sanjiva suggesting if we enforce ==1 interface per service, then the bindings could be comparably simplified
... Oracle\Umit notes loss of functionality and unclear on what scenarios it enables.
... IBM\Sanjiva agrees w/ loss of function and equates to simplicity.
... IBM\Sanjiva notes we could define a construct for multiple port types and/or to indicate relationship between endpoints (nee ports)
... IBM\Arthur notes three values:
... (a) conceptually tightens up semantics of a service
... a single instance with >= 1 protocol for access
... this is a logical extension of the interpretation from WSDL 1.1
... (b) simplifies code generation
... (c) Searching for a service that implements a given interface, e.g. in UDDI, becomes simpler if the service has an interface explicitly associated with it. I believe this is important for GRID.
... Oracle\Umit agreed that service was a bag of endpoints that didn't map well to a single programming construct.
(Why wouldn't you define > 1 interface and then a class that derives from the interfaces?)
... IBM\Arthur notes some of the desired flexibility can be met through wsdl:interface inheritance
... Oracle\Umit concerned about excluded scenarios.
... Tibco\Amy concerned that using wsdl:inheritance to aggregate endpoints is not a simplification.
... (c) simplify search through directories ala UDDI or GRID
... index a wsdl:service on the basis of its single wsdl:interface
... Macromedia\Glen views status quo
... multiple endpoints bound to disperate interfaces
... as complex and difficult to reason about
... and does not map to intuitive definition of 'service'
... interfaces are important enough to be implemented by all endpoints within a service,
... or could be implemented within another wsdl:service
... Oracle\Umit would prefer an explicit mechanism rather than opaque relationship between separate wsdl:services or having to use wsdl:interface inheritance
... could identify an instance that could host multiple wsdl:services
... Microsoft\Gudge noted that UDDI supports storage of WSDL 1.1
... and WSDL 1.1 supports wsdl:service with > 1 wsdl:portType
... and problem must be solved
<umit> It is more than a preference. Currently the implicit semantics that is associated with
<umit> service to be an instance and containing multiple portTypes
<umit> should be retained.
... Grid\Steve believes ==1 wsdl:interface per wsdl:service would simplify UDDI client behavior
... would not have to drill through structures
... expect WSDL 1.2 tModel could be simpler -- more direct
... Macromedia\Glen gave example of multiple, related departments
... but some clients would like to be able to ignore relationship
... IBM\Arthur noted we don't have a first-class way to indicate relationship between endpoints
... BEA\David proposed solution on e-mail
* alewis wants single-instance service semantics.
<sanjiva> big +1 to what Tom is saying
... Macromedia\Tom concerned about new constructs
<sanjiva> if we can't get consensus to simplify then I'm against creating new constructs!!
... in favor of ==1 wsdl:interface per wsdl:service
... called to finish up what we have and make it work well
... Chair\Marsh asked if we kept the status quo, what would we want to clarify?
... Tibco\Amy a wsdl:service is a single instance of a Web service
... a single state machine, a single instance
... (not cross organizations)
... Grid\Steve not comfortable with tying wsdl:service to Web service instance
... Macromedia\Glen: Should we make clear that a <service> represents a single entity (i.e., ultimate destination node in SOAP parlance)?
... Grid\Steve asked about implementations that do not use SOAP
... IBM\Sanjiva agreed with Macromedia\Tom that if we do not simplify wsdl:service, then we should not create a new construct
... Macromedia\Tom reiterated in favor of ==1 wsdl:interface per wsdl:service but not in favor of new wsdl:super-service construct
... Chair\Marsh noted new construct could be candidate for extensibility
... Macromedia\Tom noted proposed limitation to single entity is not a testable assertion
... Chair\Marsh expects to discuss this at the face-to-face meeting
--------------------------------------------------------------------
7. R085 Describing endpoint references. [.1]
[.1] http://lists.w3.org/Archives/Public/www-ws-desc/2003Apr/att-0088/R085-20
03-04-22.html
... IBM\Arthur noted significant discussion by MB opposed to addressing via any other than URI
... also a post from Serge re: similar proposal that includes XPath to identify which portion of the message contained the reference
... minimal discussion by WS Desc members
... Chair\Marsh wondered if a reference could be described as xs:anyURI
... and wondered how one would indicate that the reference could be an endpoint reference to a given wsdl:interface
... IBM\Arthur noted that if you include binding information in the schema, then the wsdl:interface isn't reusable
... this implies the binding information is at the binding level
... and the endpoint information needs to be treated as part of the wsdl:interface at the 'abstract' level
... and defer binding information to the binding level
... IBM\Arthur investigating using XPath (looking favorable) or Schema component descriptors (not looking as favorable)
... Chair\Marsh encouraged e-mail discussion and expects to discuss at face-to-face
--------------------------------------------------------------------
Summary of new action items:
ACTION: Sun\Roberto to write up uniqueness problem in Schema
ACTION: Marsh to summarize WS Desc interest for XML Schema subset to XML Schema WG
ACTION: Editors to clarify that recursive interface inheritance is disallowed.
Received on Wednesday, 7 May 2003 14:40:49 UTC