Minutes of Telecon 22/Oct/13

AGENDA

1. Release readiness and coordination (*Andrew*).

        - Presumably not incl. "int" support in GetHelper? Presumably convert to AI

        - Summary of exact process and responsibilities

           Date of packaging tars.

           Who packages CPP and Java.

        Testing - Who is expected to test the tar files.

        Date of announcement.

2. Release announcement. Need to collect input. Who drafts email. Who posts to tech-talk.  

3. Synopsis of Python discussion (*Marty* and *David*)

4. Windows Port (*Andrew*)

     Status of review

        Peter's questions - mainly shall we allow VS files in the repo. Is there a way we

        can support IDEs etc.

     How is Peter's port associated with PSI's CosyLab contract?

5. Modifications for array copy efficiency (*Marty*,*Andrew*)

  There are many questions raised in Marty's doc. Let's please do main comments in email,

  then review questions in next week's telecon. However, let's get started on these

  two in the meeting this week:

     - RESOLUTION that the next version will require 3.15

MINUTES

Present: AJ, BD, DH, GW, MK, MS, TK

Scribe: BD

Chair: GW

Release Readiness: MD MS MK all say they are ready for the release

AI: David will update all version numbers in RELEASE_VERSIONS file for both C++ and Java, and notify Greg.  They will then generate tar files (David C++, Greg Java).

RESOLUTION:

Updating RELEASE_VERSIONS and tar packaging will be today 22/Oct/13

Uploading to SF today

AI: Greg (Java) and Andrew (C++) will verify the tar files from SourceForge

Date of announcement:

AI on Andrew, Thu 24, Draft release announcement.

RESOLUTION: Release announcement date: Fri 25th for EPICS 4.3 Java and C++.

NEW ITEM: Summary of Python discussion from ICALEPCS

AI: Andrew will forward transcription of his notes of the discussion that took place at ICALEPCS.

Largely the discussion was orinted around whether the Python interface would be a python

implementation of pvData, or a “wrapper” to a ctypes implementation. pvAccess would

be reimplemented in python.

AI: Bob to pin down how badly/when he needs a python binding.

We’ll try to assign resources after we know what the urgency of the need.

New Topic: Windows Port

Peter Heesterman, from JET (UK).

  - Has made a large number of small modifications.

AJ: … Most changes are DLL import/export …

MS: Points out that PSI requires a Windows services side.

MS does intend to merge Peter’s changes into pvDataCPP, pvAccessCPP and pvaSrv, and to

fix any items he finds outstanding.

MD: Are we talking about any more than DLL (Dynamic libs) building. If not, then just build statically.

AJ: Thinks not, PH’s code probably mostly oriented towards DLLs.

AJ and MD feel that Peter’s code should be reviewed BEFORE it is merged into pvAccess and pvData - even by Matej.

AJ speculates that hg bundle can be read to evaluate the changes proposed by PH. But MD says output of hg bundle is binary.

MD thinks very difficult to just look at a diff. MS++

MS: recommends that he evaluates PH’s changes by eye, and reports back.

AI on MS by Nov 5th: Report on PH’s Window’s port changes and evaluation of work required.

In principle we will adopt Peter’s changes.

MD: If we will add Windows to the official supported list, then who will be our official tester?

TK: PSI can be the official tester.

MD: A real name.

AI on TK by 29/Oct:13: Make sure PSI can assign a real name of a person to be the official Windows port tester.

NEW TOPIC: Modifications for array copy efficiency

MK: It will be difficult to support both 3.14 and 3.15

TK: If we require 3.15 there will be resistance from PSI