EPICS Version 4 Telecon, 9-Feb-2016


  1. AIs status [1] and pull requests [2]
  2. pvAccessCPP #23, If more than one provider is registered then attempts to connect to any channel not provided by the first provider fails [3]
  3. Release process streamline, lessons from the Patch release (Dave) [4]
  4. Lund preparations

[1] Previous Meeting: https://docs.google.com/document/d/15iOasweS4HM-9HuqQJQodj7lYUmA2IAxI32cdkp7G6M/edit#heading=h.kjxyqj1qmjoa

[2] Open Pull Requests, https://github.com/pulls?utf8=true&q=is%3Aopen+is%3Apr+user%3Aepics-base+

[3] https://github.com/epics-base/pvAccessCPP/issues/23#issuecomment-181509750

[4] https://sourceforge.net/p/epics-pvdata/mailman/message/34834107/


Present: MS, AJ, MD, GS, DH

Scribe: AJ

Chair: AJ

GW: Hi guys, seems I can’t connect to the hangout. Other internet things ok!

Andrew - can you take over ?

AJ: Hi Greg, No David (no Internet at Diamond), no Marty.

GW, still no ability to join. Mac never connects. iPad connects and then crashes

AJ 2 GW: Where do you get the AI spreadsheet from?

GW: Ralph or I copy and paste from the previous minutes. Normally Ralph does it. I didn’t last night!

See Previous meeting [1]

AJ: Pasted in below…

NEW TOPIC : AIs status [1] and pull requests [2]

GW: #2 dbGroup funding still in progress.

GW: #10 Will do literature page v soon now other pages have settled.

DH: #12 completed, just needs announcing

AI on AJ: Discuss with GW and announce release.

GW: If Dave’s happy, then it seems ok to me to announce. It’s a patch release right? So maybe we don’t need the full announcement hullabaloo? Just asking. (AJ: OK, will do)

MD: #16 change action item, will commit directly since it’s simple. Some still to do.

Pull requests

MD: Ask DH to look at pvDataCPP#17.

DH: Was looking at it. Doxygen for serialize command?

MD: I put in the comments by habit.

DH: Some things missing? @parameter etc? deserialize from vector not done yet. Need time to finish looking at it.

MD: Will add that.

MD: header file moves should be merged soon to prevent conflicts.

DH: NormativeTypes not urgent, will do pvDataCPP soon. AJ made comments about code reorg.

MD: Will add code reorg to the normativeTypes pull request.

DH: Where’s Ralph?

AJ: On vacation this week, can’t attend telecon for the next few weeks.

DH: I’m waiting for him to merge my pull requests.

NEW TOPIC: pvAccessCPP #23

MS: <describes issue> Channel Security provider must support all channels through the provider it is registered with. Marty registered both CA and local provider under the same server.

MD: Provider is tied to the channel, security provider is tied to the connection. Security provider design needs to be overhauled, no distinction between authentication and authorization.

MS: This is there

MD: Works if there’s only one provider. This design may have performance impact because the security channel interface uses ID numbers, not pointers.

AJ: What to do?

MD: Change the interface, redesign soon. MS, MD to work on.

AJ: Also ask ESS, they will need it soon.

MD: Wire looks OK, but API doesn’t seem to support changing auth during subscription.

AI on MD and MS: Discuss and come up with proposal on redesign (+ prototype) for the auth* interface.

New Topic: pvAccessCPP #24

MS: Bug that MD found in the (de)serialization code with big-endian. Java uses big-endian, but we haven’t seen it yet.

MD: Only hits when bits between 56 & 63 are set. Issue includes link to my test code.

MS: Fixing this bug would change behaviour

MD: Only for big endian systems, little-endian systems won’t change.

GW: In what sense would it “change behaviour” (for big-endian)?

MD: It will fix the bug.

GW: Well, I was hoping not to have to write also “apart from fixing the bug”!

MD: For big-endian systems (including Java) this is an on-the-wire change, for little-endian it isn’t. Java server + C++ little-endian client can’t communication at the moment if the 56-63rd items in a structure (bitset) change, the bit-mask will be wrong.

AI on MS: Check specification, change the code to match unless the consequences are dire.

MS: Talked to SNS (Klemen) regarding reported missing field bug, it’s a threading issue. Reading in the same thread works, so this is probably a bug in their code. Swapping arrays their threads get the wrong array or something like that.

MD: Locking issue probably, they may be sharing data structure without realizing it.

MS: On the wire looks OK.

NEW TOPIC: Release process streamline, lessons from the Patch release

DH: See email. Relatively easy to do it all myself. Discovered we weren’t actually packaging the Release Notes into the Java bundle, now moved into bundleJava. Scripts are taken off master on pvDataWWW repo, not off a branch. Need a separate module for the packaging scripts; branch and tag the bundle module as well as doing that for the code modules. Split RELEASE_VERSIONS into two inside the bundle modules, one for Java, one for C++.

GW: This all seems sensible. Just a note about the “bundle” repo name - called bundleJava it sounds like it’s a complete packaging of the Java modules. Same for C++. Better name?

DH: BundleJava already exists with that name…

GW: Oh, ok.

NEW TOPIC: Lund Preparations

AJ: No Timo, I’m still waiting for hotel details; can someone prompt Timo

GW: I’ll see him today. Will do.

AJ: Greg - Anything else about Lund?

GW: Just wanted to get a sense from Dave and Marty that they (Dave in particular) is going well in materials prep and can approach other lecturers soon if needed.

DH: Not worked on notes recently, will resume soon.

AJ: Will you want help with the lectures, notes?

DH: Re Prep: I will create content, volunteers to improve them. Question about one room or multiple, will we have other people to walk round & help. For now I will work on content and see where we go from there.

GS: FRIB has a conflict and won’t be going.

GW: I think limiting to a single room is sensible- otherwise stretched too thin.

GW: Will get Timo to make sure a room of 25 (?) is available.

AJ: Can we get Timo to attend telecon next week to answer questions?

GW: Will ask. Also documented conversation (email) good for these things, since he’ll be asking other people to do the work.

AJ; Thanks.

Any other business?

GW: Is Dave going to see through item on release bundling - execute recommendations in his email?

DH: May need help from Ralph for Cloudbees, start with working out what the name. Not top priority, just took a big chunk for this release. So yes, but not immediately.

GW: Sure. Lund prep is probably more immediate. My experience - 40hrs / 1 hr needed for prep!

AJ: Ack.  

Anything more?

Lunchtime! Dinnertime, Coffeetime etc.

Ciao everyone.

Thanks Greg, will you publish these?

ok sure.

Active Action Items

Note: Completed AIs are shown once in the list of the meeting in which they are recognized as completed. They are removed in the minutes of the following meeting.




Target Date



See who can fund dbGroup




Write the one license file (using the MIT license in most cases)




Update your module to include copy of that one LICENSE. Update your headers to contain the header text.


wait on 5


On your module pages, remove the current/previous links; replace “Working Draft” with version number of module, which has its own numbering scheme, like 5.0.2 for pvDataCPP. Eg “Version 5.0.2 18/Nov/2015”





Simplify Literature page and home page to help users find novice oriented documentation.




Usher the patch fix for buffer overflow through to release.





Commit changes to all modules, other than exampleCPP and pvDatabaseCPP, which set the shared library versions