EPICS V4 TELECON, 25-MAR-2014

AGENDA

0. Preliminaries:
  New meeting time. Proposal for 1/2 hr earlier.
1. Why so many "failing" builds? (10 mins)
2. Review Marty's "proposed changes" and associated AIs [2] [3] [4] (30 mins)
3. Lock-up in pvAccess (10 mins).
  What was this issue, why will a codec version fix it?
4. NTNDArray proposal review, and Andrew's comments [5] (10 mins)
5. Early thinking on next release (if time, may be next week)
  What will be in it?
   - wait for pvGrp from Ralph?
   - pvDatabase included or not?
   - Bundle pvaPy?
   - Bundle matlab wrappers?

[1] http://epics-pvdata.sourceforge.net/home.html

[2] https://sourceforge.net/p/epics-pvdata/mailman/message/32123768/

[3] http://epics-pvdata.sourceforge.net/internal/proposedChanges/proposedChanges_3_0.html

[4] Diff of previous proposedChanges_3_0 to this new one - they are extensive,

http://services.w3.org/htmldiff?doc1=http%3A%2F%2Fepics-pvdata.sourceforge.net%2Finternal%2FproposedChanges%2FproposedChanges_3_0_20131023.html&doc2=http%3A%2F%2Fepics-pvdata.sourceforge.net%2Finternal%2FproposedChanges%2FproposedChanges_3_0.html

[5] https://sourceforge.net/p/epics-pvdata/mailman/message/32097442/

[6] https://sourceforge.net/p/epics-pvdata/mailman/message/32126946/

MINUTES

PRESENT:  AJ, SV, MK, GW, DH, MS

SCRIBE: MS

CHAIR: GW

APOLOGIES: TK,RL

Check with Bob on new meeting time,

Preliminaries:

- May have to revisit meeting time for Bob and Matej

******

NEW TOPIC: Why so many failing builds

AJ: pvDatabase has bug in example in

You need to adjust the release file in the exampleServer subTop

AJ: Windows problems are in name decorations.

MS: No decorations for pvDatabase at all

AJ: Not sure why pvaSrv-win64 jobs are failing. There is a test failing in pvAccess [MS understands].

AJ: Can MS take a look at the errors in the build serve [for pvaSrv Win64 and 64s]. May be c++ issue.

******

NEW TOPIC: Review Marty's "proposed changes" and associated AIs

GW: Should a cookbook be in general module?

MS: Cookbook is only about pvData.

GW: It needs to be published automatically by Jenkins build.

AJ: There is a script in cloudbees subdirectory that does sync. It should be possible to use this script.

MS: It’s a very short text file.

AI on MS: Matej makes cookbook automatically published by Jenkins.

AI on MD: Review the documentation about shared_vector in present pvDataCPP.html and new implementation of PVArrays

ISSUE: MS and MK do not have Windows, vxWorks or RTEMS themselves for development, so when there is a build issue they can’t easily fix it.

AJ will closely observe and help with problems in Windows, vxWork and RTEMS.

In general we agree that responsibility for a module’s health rests with the module owner but if another programmer breaks it they have responsibility to fix it.

AJ: Currently default branch is inactive.

MK: I need to solve this.

AI on MS: Fix pvDataCPP repository to have a default branch back (not “inactive”).

 

GW: We need to define standard for a syntax for using const.

MS: No need, use C++ books.

AI on DH: Implement changes w.r.t. const (proposed changes 7.1)

MK: I not remove “deprecated” method in Java, since pvIOCJava depends on them (XML parsers). And I use pvIOCJava for testing.

GW: Can we deprecate pvIOCJava (remove pvIOCJava from next build) for some time.

AI on MK: pvField Methods (7.2) will be removed from CPP modules, however they will only be marked deprecated in Java.

AI on MS: Make the changes to pvAccess API in both Java and C++.

AI on MK: Make changes in pvDatabaseCPP, pvIOCJava.

AI on RL: make pvAccessAPI changes w.r.t. proposed changes (Sec 10) in pvaSrv.

Ai on MK: Review and make changes to pvDatabaseCPP to pvCopy and local channel provider per Proposed Changes sec 13.

******

NEW TOPIC: NTNDArray proposal review, and Andrew's comments

http://epics-pvdata.sourceforge.net/alpha/normativeTypes/normativeTypesNDArray.html

DH: Did not renamed “reverse” to “reversed” since the equiv is named reverse in areaDetector.

Did make it boolean. Binning is still an int.

AJ: So binning tells you how it’s binned, not whether binned or not?

DH: correct. Conventionally 1 means there is no binning.

MS: Should I add NTNDArray to eget and CSS (Vtypes)?

MK: Yes.

GW: Don’t want MS to work on CSS too much.

MK: Should GW email BD to ask GC to work on this?

MS: I can create mapping for Vtype, done this before.

MS does need a vtype from Gabrielle to which to map, so requires input and action from Gabriele.

AI: on GW write BD on priority of NTNDArray support in CSS (v-types).