AGENDA
=======

agenda is to review the agenda of the Jan 2013 PSI meeting, draft below

1st draft agenda for comments.
 
Monday
======

Morning - Demos of Beta 2 and administration

  * Demos of bpm orbit and model services (*Timo*, *Greg*)
  * Demo all normative types to pvget and eget (*Matej*)
  * 2012-2013 charter review (*Greg*)

Afternoon - EPICS V4 control database & IOC requirements and design references

  * Marty's review of the existing pvDatabase, 
    [Please forward title to greg] *Marty*.
  * Marty's Ideas on memory, multithreading and locking from Marty's 
    email of Dec 12 *Marty*.
  * Bob's review of V4 IOC requirements 
    [Please forward talk title to greg] *Andrew Johnson*
  * Andrew's review of V4 IOC requirements 
    [Please forward talk title to greg] *Bob Dalesio*
  * Michael's review of V4 IOC requirements 
    [Please forward talk title to greg] 
        *Michael Davidsaver or substitute*
 
Tuesday
=======

Morning - Beta 2 continued

  * Microbenchmark testbench design and initial performance 
    breakdown report (*Matej*)
  * v3Channel review (*Marty*. Summary of the converation and outcomes from 
    Marty and Andrew's discussion around mid november in the thread "multichannel puts")
    Status breaking v3Channel out of pvIOCCPP (*Ralph*) and 
    testing by Andrew.
  * Impressions of getting started with vxWorks and v3Channel 
    compilation *Dirk Zimock* 

Afternoon - IOC image processing requirements and support plans

  * Make a real plan for the compression support of NTImage. 
    Who can talk to this, *Nikolay?*
  * NDarray / image server - David?

  What else can move us along here?


Wednesday
=========

Talk to beamline specialists at PSI. Timo and Bob to coordinate.

MINUTES
=======
Present: BD, TK, MK, RL, MS, GS, GW, NM, DH
Scribe: GW

*******
NEW TOPIC: Use of CI servics 
*******
RL: There is now cloudbees [for Java]
RL: For Gabriele, it may be that c/c++ compilers can be supported. Will not be easily possible to create the full env for compiling EPICS stuff or vxWorks cross-builds though, since it's only linux (F17) on the amazon cloud.
RL: We will need the EPICS base (makefile system, libraries) for instance to do our CI. RL can try to install EPICS base, but does't know.
GW: Seems hard for linux c/c++ and since impossible for vxWorks how about BNL's Jenkins
RL: Gabriele was the support person at Jenkins, and the Jenkins BNL machine is probably going to die.
GS: Irmis is also on SF. In summary, jenkins not needed for IRMIS at BNL any more, so Jenkins is not going to be supported any more.
MS: I have an account, and have set up projects on BNL's Jenkins, so if it's available, then I can configure it. However I would also need some login access.
GW: Can it be that the BNL system manager takes care of sys admin, and MS takes care of config.
BD: We can't support that service anymore.
MS: We can ask CosyLab to set it up

AI on MS: Get CosyLab to make a Jenkins server available for EPICS V4, and configure it for Java and Linux C++ builds. BNL will pay.

MS: Do we have an official Linix distro target?
GW: no
RL: Maybe target Debian or RHEL to be able to use binary installs for EPICS base and such. Maybe best have 2 machines, one for 32 bit and one for 64.

GW: Would like vxWorks "port"
TK: PSI has vxWorks licenses. In the meeting, we can try to convince Dirk to do the vxWorks CI.

-- Add vxWorks CI and other matters to the agenda --.


RL is also going to try to set up the C++ support of CI for EPICS V4 core on Cloudbees. 
However, it won't be able to do multi-machine or cross-platform client-server tests
MS: Well, we need client-server tests, that's what's most important.
RL: I can do client-server if I use localhost. 
RL: The client server tests of the Java builds [of the core pv modules] are working now. 
We just have one test that is failing - issue has been raised in SF tracker.

*******
NEW TOPIC: Confirming Agenda for PSI meeting 21-23 Jan
*******

Morning - Demos of Beta 2 and administration

  * Demos of bpm orbit and model services (*Timo*, *Greg*)
    TK says yes he'll try.
  * Demo all normative types to pvget and eget (*Matej*)
  * 2012-2013 charter review (*Greg*)

Afternoon - EPICS V4 control database & IOC requirements and design references

  * Marty's review of the existing pvDatabase, 
    [Please forward title to greg] *Marty*.
    MK
I will not attend the upcoming workshop at PSI.
I am having more and more difficulty walking.
This Saturday I will have an MRI on my spine and brain.
Next week Tuesday I will have a followup exam with my neurologist.
Maybe then I can find out what is causing my problems.

  * Marty's Ideas on memory, multithreading and locking from Marty's 
    email of Dec 12 *Marty*.
    pvDataCPP

Michael Davidsaver:
1) has pointed out that the existing implementation of PVArrayData results in an array copy evertime PVValueArray::get is callled.
2) has convinced me to consider modeling  PVValueArray as a C++ STL container rather than following  the Java model.
Look at repository pvDataCPP-md to see his work in progress.
  * Bob will talk on broad matters of high throughput large dataset.
  * Andrew's review of V4 IOC requirements 
    [Please forward talk title to greg] *Bob Dalesio*
  * Michael's review of V4 IOC requirements 
    [Please forward talk title to greg] 
        *Michael Davidsaver or substitute*
 
 BD: Are we going to meet on Thurs before?
 TK: Mainly on Wed
 BD: Is Marcus goign to be there
 TK: At least on Wednesday. I am going to invite SwissFEL
 BD: Imageserver, multidimentional array service, ar what we want to talk to them about.
 BD: Is there some time where we can talk to SwissFEL abotu what we plan to do for beamlines?
 GW/TK: Yes absolutely.
 BD: I can give a talk on 
 
 Tuesday
=======

Morning - Beta 2 continued

  * Microbenchmark testbench design and initial performance 
    breakdown report (*Matej*). yes
    
  * v3Channel review (*Marty*. Summary of the converation and outcomes from 
    Marty and Andrew's discussion around mid november in the thread "multichannel puts")
       From pvIOCCPP.html

It  also provides the ability to get and put a set of V3 records. Only data  from field VAL is allowed and all the V3 records must have the same  scalar numeric type. The data appears as a scalarArray. The records can  be processed. If all V3 records are in the same lockSet then each get or  put is an atomic operation, i. e. dbScanLock is called only once. If  the records are not in the same lockSet each record is individually  locked.

    Status breaking v3Channel out of pvIOCCPP (*Ralph*) and 
    testing by Andrew, plus Demo.
 
  * Impressions of getting started with vxWorks and v3Channel 
    compilation *Dirk Zimoch* 

Afternoon - IOC image processing requirements and support plans

  * Make a real plan for the compression support of NTImage. 
    Who can talk to this, *Nikolay?* 
  * NDarray / image server - David? yes. one serves periodic image
    Possibly also that as an areaDetector plugin! That is, NDarray inserted into the image.
    Demo included.
    
  * CSS and PvManager status (Gabriele). Ralph will manage connection to Gabriele.
  * Plans for CSS/pvAccess integation.
    
      

Wednesday
=========

Morning
------------
Talk to beamline specialists at PSI. Timo and Bob to coordinate.
Bob is going to give a talk on BNL's - sending it to me.

Afternoon
--------------
Versioning, release procedure(Ralph)