Minutes for twentyfirst OGSI-WG Teleconference
February 11, 2003 @ 16:30 - 18:30 GMT

Dave S.
Karl C.
Steve G.
Steve T.


1) Approve minutes of February 5th, 7th, 10th Teleconferences - Skipped.

1a) Bug Contradictions

   Bug 49 Says all but Sink MUST be extensions GSPT. The text is ok, but not the
          draft wsdl. We need an action.

Action: ---: Feb 11: Dave S: Make PT extensions explicit in the text. - Done.

Action: ---: Feb 13: Tom M: Make PT extensions explicit in the wsdl.

   Bug 53 Resolves query by service data name to have both singular and plural
          forms and the text is arguing about what the single plural name should be.

   Bug 92 Resolves that we use queryByMultipleServiceDataNames consistently, and
          make sure the singleton reference isn't there.

Resolved: Only a single operation ending in '...Names' without the word 'multiple'.
Action: Dave - Done.

   Bug 92 Resolves we adopt the name list in attribute syntax for
          find/set/delete/etc/ServiceDataByName.

This was a misunderstanding, question dropped.

Other service data bits:

9.1:

the serviceDataName SDE set must contain all SDEs included in the WSDL and MAY contain dynamically added SDE names.

Action: Dave - Done.


9.2.1.1:

The name listed in the query MUST be among the serviceDataName SDE (see 9.1) contained in this service instance.

Action: Dave - Done.

   Bug 78 Says life time attributes need to be on GSRs. They aren't.

This was a misunderstanding.

   Bug 86 Says: "We have agreed to put language to the effect that the semantics
                 of all SD SHOULD be included in the service's description. We
                 should do so for the SD in the spec. Therefore, we need to
                 include coherency descriptions of all the Service Data that we
                 define in the spec." - Who's bright idea was this? ;-)
Action: Steve T.

   Bug 91 Says use minOccurs=1 and nillable rather than minOccurs=0 and explain what
          nil means. This has not been done for:

Action: Dave to try a pass. I did as below:

serviceDataName: Actually minOccurs=6. These 6 can't be nil.  nillable=�alse�

FactoryHandle: minOccurs=1, nil => not created by a Factory, text was OK.

GridServiceHandle: Leave as minOccurs=0, as this make sense for a service not advertising itself, or not knowing what 
handle scheme it is under. Nil has no meaning.  nillable=�alse�

GridServiceReference?s?: The text implies that minOccurs=1. And nil has no meaning.  nillable=�alse�

NotifiableServiceDataName: Leave as minOccurs=0 as this makes sense.

Entry: MinOccurs=0 Makes sense. Nil has no value. nillable=�alse�

Note: I made the  nillable=�alse�explicit. This is, I believe the default, but it can't hurt.


1b) Filter of add requests on ServiceGroup

Two aspects: 1) SG Property "I have only members of PT X or Y or ...". Add fails
             2) Implicit filter on findServiceData

Dropped as being the same issue as with UniquenessProperty. Can be affected by the subtype.

1c) The Pen

Pen to Steve T soon. Steve To hold until Wednesday's call.

2) Review Faults Section

Meta Issue 1:

From Steve T: The biggest comment/issue that I see is that our model of having each fault
              reflected in a WSDL fault element in the operation declaration won't work
              for extensible operations such as FindServiceData. Should we go back to having
              just a single fault in the WSDL operation of type GridServiceFaultType, which
              is extended as necessary?

Deferred discussion until later.

Meta Issue 2:

Resolved: Defer the fault hierarchy until v2.
Action Steve T to effect changes to spec.

Questions:

   a- Multiple errors in an operation? - Dave S would recommend that the "first" error be the
      primary error & others form part of the extensibility element, which may be a sequence
      of GridServiceFaults. Not this is not the same thing as the "cause" element.

Drop question. A single fault is returned and...