Present: BD, GS, MS, MK, GW, JR
Scribe: GW


NEW TOPIC: Use of SWT in pvIOC distribution.

RESOLUTION: We will move the pvioc.swtshell package to its own EPICS V4 project [that is, a new independent repo].

Marty and Matej will do that either before or after the BNL Oct Meeting, depending on their own progress at meeting
the demo objectives from the September 5 telecon (http://epicsv4.titanpad.com/ep/pad/view/27/latest) .

NEW TOPIC: policy for choice of wire provider (v3Channel or CAV3) when both are possible.

AI: Matej will develop a policy that implements the mechanism to choose. One policy choice will be that
both providers will be given a chance to respond and a preferred provider will be used if it responds.

NEW TOPIC: Status of VxWorks port

AI: Matej, check with Dirk.

NEW TOPIC: ServiceClient, ServiceClientFactory and ServiceClientRequester 

RESOLUTION: ServiceClient, ServiceClientFactory and ServiceClientRequester  will be moved to pvaccess from pvService.

NEW TOPIC: BD asks Question: Is there consensus on Self-ID and why we need it? 

MS: Self ID isn't needed for pvAccess itself
BD: Is it necessary for memory allocation at an endpoint? Presumably not.
Correct.
GW: Self-ID is to help a client determine how to interpret the data in a (pvData) structure. The reason we had so much discussion about it was that it was being proposed that EPICS V4 should define a type system, and that type system should be polymorphic. Example: we define NTTable. Some institution decides to extend NTTable to produce "BPMOrbitTable" by adding some requried fields. A client that knows NTTable, but doesn't speocically know BPMOrbitTable, should be able to display those facets of BPMOrbitTable object that conform to NTTable. However, we have decided as a group that to support this, possibly multiple inheritance, system all from inside code, is too hard for this charter.