Minutes of the Ninth OGSI-WG Teleconference 09/10/2002 @ 15:00-16:30 GMT Attendees: Tim Banks Karl Czajkowski Shel Finkelstein Steve Graham Bill Horn John Karpovich Fred Maciel Andreas Savva Frank Siebenlist David Snelling Steve Teucke Peter Vanderbilt 1) Approval of minutes form October 2nd. OK 2) Review of plans for GGF, The following topics have identified and an initial priority assigned to some: 1 - Notification possibly divided into "Event-style mechanisms". 2 - Registration/Collection interface hierarchy and operations. 3 - Read-Write SDE's Bug 51 ? - Extension relationship in the GS portTypes. ? - Coherency semantics of multiple findServiceData operations. ? - Bugizlla issue review. ? - Interim strategy discussion (what to do while we wait for WSDL 1.2) ? - Resolver behavior and the semantics of multiple naming schemes. 3) Matters arising - We agreed to continue the teleconferences following GGF6, to to give ourselves a week off before resuming. Time to be agreed at GGF6 (due to the end of daylight savings). - Note to folks wanting to reread the draft in the absence of change bars. Most substantial changes have bee to sections 4 and 6. Otherwise, check the change log. If changes are made before GGF, there will be a tracked version posted. - Note that the GSS now meets GGF copyright and IPR rules. Well done all those pushing on this. 4) Technical Agenda, Service Data Issues: Confirmed that bugs 21 and 22 had been resolved. ------ Bug 48 - SDEs that should not be initialized in the WSDL It was resolved that an Enum should be added to indicate that values MUST NOT appear in the WSDL or values MAY appear. The Enum is intentional to allow for more structure later. ------ Bug 53 - queryByServiceDataName to allow multiple name parameters It was quickly agreed that this was a good idea, but a discussion followed on the semantics. In particular, what coherency semantics can/ought to be assumed between the members of the serviceDataSet returned? Given that the serviceData itself has no guarantees attached to, only coherency within the service data itself was thought to be possible. The goodfrom/until annotations within each SDE provide a 'soft' coherency relationship between the serviceData state and the instance state. Language to be sought that allows the goodfrom/until attributes of the returned serviceDataSet itself to reflect this 'soft' coherency relationship for the members of the set. ------ Bug 32 - xsd:float use is misleading As data types other than strings are only used in examples, it was agreed that these be adapted to provide their own typing information (i.e. correct the XML). The proposal in this bug is to be dropped. ------ Bug 45 - serviceData type attribute is redundant Resolved as proposed and therefore to use the xsi type approach. ------ Bug 40 - Does minOccurs=1 make sense in light of authorization? Resolved to leave it in as an indication of designer intent. The text should be checked to cover authorization caveat. ------ Bug 46 - Proper interpretation of Mutability Bug 47 - Clarify terminology of serviceData These have both been resolved and fixed in the document, see action below. ------ Bug 54 - We need some way of specifying 'nothing' as an SDE in itself. Resolved as proposed. An emptyElement type will be provided for extensibility elements that are supposed to mean 'nothing' is here. Still outstanding: Bug 34 - Model FindServiceData() & CreateService() specializations as methods Bug 51 - Read-Write SDEs - deferred to GGF6 face to face. 5) Time of meetings after GGF6 - deferred to GGF6 6) Review of actions: Action: Steve Graham to drive the creation of an XML Schema for the specification, #35 - Pending, but may be dealt with in a breakout at GGF6. Action: Steve Graham to craft language for recommendations to developers for use of resolvers. Pending, but may be a session at GGF6. Action: Dave Snelling and Andrew Grimshaw to look at requirements and solutions for pull like semantics - Pending with ideas forming. Action: Dave will look into a DeliverWithAck option. One issue is specifying which delivery operation should be used - Pending, but ties into the events and notifications discussions. Action: Steve Tuecke will propose a binding layer project to OGSA WG, but he is not committed to drive it - Pending GGF Action: Proponents of the inclusion of QoS specifications in Grid Service portTypes to produce documentation to the list justifying a reconsideration of this issue - Pending activity on the list. New Actions: Action: Steve T. to arrange calls starting October 30th at the new time. Action Steve T. to update document schedule on the Web site to reflect decisions at the interim meeting. Action: Authors to ensure that changes have been executed in the draft for bugs 21 and 22. Action: Steve G. to check bug 46 and 47 status in the document. Action: Dave to go through the minute and append resolutions to the notes associated each bug. Action: Dave and Steve T. to establish a process for dispatching the bugs and removing their references from the document. 7) Open Issues being carried: - Track the IPR status of WSIL. - Should WS-Inspection document (or other formats) be explicitly noted in the spec? - Oinn's report on presence or absence of SDEs still be be discussed. - What assumptions can be made about the properties of a service resolved by a resolver? #43 The specialization of the DeliverNotification operation #44 Language linking notification subscriptions to SDEs. - Explore further the implications of subscriptions as Grid Services. - What does a successful return from a RegisterService call mean? #42 The distinction between 'void return' and 'no return' in some operations #34 Model FindServiceData() & CreateService() specializations as methods See you at GGF6.