Minutes: 5 Nov 2003 WS Description FTF

Web Service Description Group
Minutes, FTF meeting 3-5 November 2003
Sunnyvale, hosted by Fujitsu.

Wednesday 5 November


 David Booth                  W3C
 Allen Brookes                Rogue Wave Software
 Roberto Chinnici             Sun Microsystems
 Glen Daniels                 Sonic Software
 Paul Downey                  British Telecommunications
 Youenn Fablet                Canon
 Tom Jordahl                  Macromedia
 Jacek Kopecky                Systinet
 Philippe Le Hégaret          W3C
 Amelia Lewis (phone)         TIBCO
 Kevin Canyang Liu            SAP
 Lily Liu                     webMethods
 Jonathan Marsh               Chair (Microsoft)
 Jeff Mischkinsky             Oracle
 Dale Moberg                  Cyclone Commerce
 Jean-Jacques Moreau (phone)  Canon
 Bijan Parsia                 University of Maryland MIND Lab
 Arthur Ryman (phone)         IBM
 Jeffrey Schlimmer            Microsoft
 Jerry Thrasher               Lexmark
 William Vambenepe            Hewlett-Packard
 Sanjiva Weerawarana          IBM
 Umit Yalcinalp               Oracle

Scribe: Lily

09:00 Publication review - motion to publish parts 1 and 2

JeffSch:   Looking at the part 2: it seems we've lost some changes in 
           part 2, may be just the titles.
[Amy is updating the titles in the spec right now.]
Roberto:   The schema seems not updated yet
JeffSch:   We manually update the schema, but publish it the same time 
           as the spec.
[The group have reached consensus to publish both part 1 and part 2.]
JeffM:     Attributes should be given a higher priority, because it is 
           so controversial. I think it would be good to get more feedback 
           on the subject
Marsh:     Attributes, inline schema will not be completely done.

09:15 Endpoint references
    - Proposal from Arthur/Umit [30]
    - Awaiting simplified proposal from Roberto/Glen
    - Motivation for R131 from DaveO [31]

 [31] http://lists.w3.org/Archives/Public/www-ws-desc/2003Sep/0181.html

[Roberto: http://lists.w3.org/Archives/Public/www-ws-desc/2003Oct/att-0345/counterproposal.html]
[Sanjiva: Arthur's proposal:
Arthur:   Disagrees with Roberto's proposal that puts all interface and 
          binding information in the XML Schema
Roberto:  There are many alternatives to solve this issue and the original 
          proposal is more powerful than necessary.
[Arthur:  From last f2f: 
          The presentation from the last f2f: 
Roberto:  We need to understand the requirements before we judge the solutions
[Arthur:  R085 is a brief summary of a long sequence of postings which 
          clearly define the requirement.]
Arthur:   R085 is clearly defined, based on a series of discussion
Amy:      Agrees with Arthur that Roberto's proposal brings bindings into 
          interfaces. That breaks the separation of abstract definition from 
[Glen:    reference = endpoint [binding] [interface]
          reference = endpoint binding [interface]
          reference = endpoint binding interface
          [] = contextual information, not serialized in the reference itself]
Umit:     People are making conclusions too early, we should wait until 
          Roberto finishes his proposal. All proposals have that same aspect.
JeffM:    is concerned with the interoperability.
[Umit:    Interoperability is achieved by agreeing on the form of the reference]
[Roberto's proposed solution is to introduce the usage of @serviceReference 
and service type.]
[Glen doesn't like the @serviceReference, prefers to have a base service type, 
and have the wsdl:service and the service marker derive from that base type.]
[Umit:    This problem is very similar to marking the schema with a media 
          type. The solution space is similar to approaches presented by 
          Phillippe at the last f2f. I think choosing an approach similar to 
          representing media-types would be beneficial.]
Sanjiva:  The document reference has nothing to do with the service reference.
[Arthur:  Requiring a complex type wrapper for URI's is inconsist the other 
          W3C specs, e.g. HTML <a @href="uri"> and XLink. some developers may 
          want to use attributes for URIs, e.g. 
          <Department ref="uri">Shipping</Department>]
Sanjava:  This breaks the separation rule: types defined by schema, and other 
          done through wsdl constructs.
Umit:     It depends on the type, URL referring to endpoints are treated 
Glen:     There may be valid cases to pass XML to the applications.
Jacek:    We don't need to differentiate service references at the schema 
          level, it should be done at the wsdl level.
Sanjiva:  Wants to solve the issue to describe a service reference in 
          interface/operation first
[Umit:    A service reference must be designated by a specific type at the 
          Schema level. The declaration should be in the types/schema.
Roberto:  In response to Jacek, the service and service reference type belong 
          to abstract definition.
[Arthur:  see also my example using <wsdl:reference>: 

10:30 Break
10:50 Endpoint references (cont)

[Marsh:   the whiteboard subcommittee is almost done deliberating...]
[Umit:    For what it is worth, here is some more discussion on this topic:
Sanjiva:  Describing the scenario on a whiteboard:
  int bar() { ... }
  int foo() {
    bar create()
Marsh:    We need to indicate that "bar" is an interface.
[Arthur:  Note: in C# struct is like XSD and interface is like WSDL!]
Umit:     "bar" should be of Object type in java world
[Umit:    Correction: The example presented here is what I call is the 
          semi-static case, when you know what bar is=an interface.]
Jacek:    We should be debating on where to put this "callable" hint in 
          wsdl for applications to use.
Glen:     In response to JeffM, people want to introduce 
          bar *foo()--service reference
Jacek:    This is not necessary at wsdl level, we can put it in the primer 
          to tell people how to use schema to handle it
JeffM:    We need more than an example in the primer.
DavidB:   It can fit into the primer if the case is less strong than what 
          JeffM prefers. It will need a new type.
JeffSch:  A note that the schema as of today does not have a global service 
Arthur:   We need to keep the abstract definition vs. binding separation.
JeffSch:  Can we put this to sleep and move on? It seems like most people 
          prefer minimum work in the spec.
Sanjiva:  We should close the issue.
JeffSch:  Somewhere in the spec we will indicate the service reference 
          usage. We will not introduce a new type.
[Group agree on making top wsdl elements global.]
RESOLVED: Make top WSDL elements global in the schema to facilitate reuse.
Glen:     likes a reference type in the spec and doesn't feel comfortable to 
          make the service ref name required on the wire.
[JeffSch: Per WG decision, made global element declarations for top-level 
          children of wsdl:definitions.]
Marsh:    Current proposal is to add a paragraph in the spec describing 
          that these references could show up on the wire; and to supply 
          some examples.
[Marsh: New issue: Should wsdl:service/@name be optional?   We don't want 
to force users to have to invent a name when service appears on the wire, 
but currently we require @name within the context of a wsdl:description.]
[JeffSch: New issue is #95]
RESOLVED: Add a paragraph in teh spec describing that <wsdl:service> can
          show up on the wire as a service reference.  Primer will have some
          examples (see Roberto's counterproposal as a basis).

12:30 Adjourn

Received on Friday, 7 November 2003 18:17:59 UTC