Minutes of the Eleventh OGSI-WG Teleconference 11/Dec/2002 @ 21:30-23:00 GMT Attendees: Karl Czajkowski Steve Graham Andreas Savva Dave Snelling Steve Teucke Pete Vanderbilt 1. Approval of Minutes - OK 2. Review actions Action: Jeff Frey to draw up a description of how to integrate the 'implicit' verses 'SDE' state dichotomy. Also look into other issues of coherency control. - Pending Action: Steve T&G to push on W3C to get what we need (portType extension without breaking encapsulation). Take ownership. - Pending Action: Tom Maguire and John R, not being happy with the mutability language, will propose new language reflecting the current meaning. - Pending Action: Dave S Faults summary. - Pending Action: Steve Graham and Pete to drive the creation of an XML Schema for the specification (#35). Dependent on service data and some pointers on WSDL 1.2 syntax. - Pending Action: Dave Snelling and Andrew Grimshaw to look at requirements and solutions for pull like semantics. - Pending Action: Dave S to put a proposition for registration portType, together with exact text. - Pending Action: Andrew G to write a document describing the options for reslover behavior. - Pending Action: Steve Graham to craft language for recommendations to developers for use of resolvers (#17) - Pending Action: Steve T. Email list of proposed resolve->verified bugs - Done. Action: Pete V. and Steve T. to propose a more general solution to the extensibility argument pattern, see bug 34. - Pending Action: Pete V. to create XML schema for locator elements, see bug 37 - Done. 3. Workgroup Interim meeting: Planned for Jan. 22-24 2003. Any one wishing to host should contact STeve T before the end of the week, or we will default to Chicago. 4. Close loose ends on mutability and modifiability: >From the list Peter V. wrote: ------ Second, "setOnce" implies that there may be a state in which the SDE is not initialized. Presumably a settable SDE is present to affect or control the service. But if it's not set, then we have a service that's not completely initialized. What does the service do if it gets a request before the SDE is set? What happens if the SDE is referenced before it is set? Is there a standard representation for an uninitialized SDE? In my opinion, this is an unnecessary complication. If the SDE is not modifiable, the service should set the value at service-creation time. If the SDE should be supplied by the client, make it a parameter to the factory's createService method. In both cases, the service is fully ready to go immediately after creation and none of these complications come into play. ------- The discussion covered the question of wether an SDE could remain unset for a time or had to be present when the service started. One proposal was to drop the setOnce concept and let it be covered by minoccuss and lifetime management. However, it was felt that the setOnce concept was rather useful, but that the model was cleaner with the above proposal. therefore: Resolved: setOnce SDE MUST be initialized before the service is accessible to clients. Therefore the table of options is now: Mutability Modifiability Comments ---------- ------------- -------- constant false Appears in the WSDL only. setOnce false Not in the WSDL, exists as soon as the service exists, better name? append true Can be used with minoccurs=0 maxoccurs =1 to do client side "setOnce" semantics. append false As before mutable true As before mutable false As before 5. Service data We had a short discussion of the Service Data document under preparation by Steve G. and company. There are several issues that will arise around this document. These were highlighted briefly in discussions. To summarize: - Generally agreed that some way of making the Service Data concept appeal to the Web Services community should be sought. - It was not so clear that making this document separate from the Grid Service Specification was the right way to go. - It was clear the there will be some discussion on the port decoration sections. - Previews of the ideas in Jeff Frey's section on service data semantics indicate that discussion will be required. - We might consider splitting off a Service data portType from the root GridService as a way of lowering the threshold to WS community. We hope to have a draft by the end of the week. 6. Confirm bug closure: The following items have been fixed and verified for some time. They will be closed after these minutes have been approved at next week's teleconference. 4 FIXE compatibilityAssertion: who is making the assertion 6 FIXE Move ServiceImplementation to a serviceData Element 7 FIXE Drop Compatibility Assertions 8 FIXE Make Handles URIs 9 FIXE Add ServiceData descriptions to a grid service description 10 FIXE Nuke PrimaryKey PortType 13 WONT Use Measured TimeStamps 14 FIXE SetTerminationTime of "forever" 15 FIXE Combine SetTerminationTime and Destroy 18 FIXE Add gsdl:subscriptionExpression type? 22 FIXE What if SDE lifetime attributes are not specified? 25 FIXE Version numbers vs immutable service description 26 FIXE Who controls handle namespaces? 28 FIXE Restrictions on http GSH resolver protocol 29 FIXE Operation to check if GSH (or two) refers to instance 33 FIXE QNames as sole extensibility mechanism 36 FIXE Does OGSA need a more reliable way to resolve qnames? 38 WONT unique service instance reference 40 FIXE Does minOccurs=1 make sense in light of authorization? 45 FIXE serviceData type attribute is redundant 47 FIXE Clarify terminology of serviceData 54 FIXE We need some way of specifying 'nothing' as an SDE ... 12 LATE Change Notification Source/Sink to Producer/Consumer? - Close as WONTFIX 27 INVA Multiple GSHs within one URI scheme to the same instance - Close as WONTFIX 64 INVA Does minoccurs >= 1 make sense? - Close as DUPLICATE 16 LATE Absolute or relative termination time - close as WONTFIX 5 WONT Factory extensibility 52 WONT Refactor Registration portType 57 WONT We need events and Topics 65 FIXE Do we need to put IPR statements in the Spec? - close as WONTFIX 7. Registration This bug can be closed as we have agreed not to include WS-Inspection at this time. 20 LATE Is WS-Inspection document required in registry? There was a short discussion about what needed to be there instead. The minimum properties requested were: - Handel/Locator - References - Service WSDL, - Most derived porttype - link to other services (Locator) This action is alread with dave S, and Steve G. has offered to help with the angle brackets. 7. Buggzilla Sweep from Bug 39. none. 8. Next Meeting: December 18, Wednesday pm (US & Europe) or Thursday am (Asia-Pacific) at: 13.30-15.00 US Pacific (e.g. Los Angeles) 14.30-16.00 US Mountain (e.g. Denver) 15.30-17.00 US Central (e.g. Chicago) 16.30-18.00 US Eastern (e.g. New York) 21.30-23.00 GMT 21.30-23.00 UK (e.g. London) 22.30-00.00 Central Europe (e.g. Paris) 06.30-08.00 Japan (next day) Teleconferences are scheduled for: January 8, 15 The teleconference information for all remaining scheduled teleconferences is: 1-484-630-8733 Passcode: 72316