Personal tools
  • We're Hiring!

You are here: Home Community Minutes Conference Calls 2015 2015-08-04 Tuesday Team Meeting

2015-08-04 Tuesday Team Meeting

Dundee: Kenny, Will, Mark C, Petr, Jean-Marie, Ola, Simon, Jason, June, Roger, Chris Remote: Sebastien B, Kelli, Ian, Helen, Andreas, Stick, Balaji, Mark H, Melissa, Josh B, Wilma

Agenda - 2:30pm Start

  1. Accepting minutes from last meeting

    • accepted
  2. Project Timelines (2-3 minutes each)

    1. OMERO - 14:30 UK

      1. Mainline (QA) 5.1.4, 5.2.0

        1. new CI jobs (Sebastien B)

          • contributing PR open describing 5.1 line on trout, DEV (5.2) line on eel, queues seem to be working
        2. breaking status

          • Sebastien B: jsTree PR (Will, Douglas) is okay on breaking, but can be moved to a space if necessary
            • Will: what are the criteria for taking it out of breaking? It’s mostly working, tests are passing.
              • Jean-Marie: list in stand-up for review; after that, remove breaking tag
                • Ola: proper testing round?
                • Jean-Marie: yes, early next week?
                • Will: that would be fine
              • Will: some outstanding to-do’s noted by Douglas
            • Sebastien B: Robot Framework tests are not run against breaking, could be done though
              • Will: can also run them locally
              • Simon: Josh has been working on CI framework in Docker, but Robot Framework isn’t yet there
        3. Major PR merged plan.

          • Sebastien B: do version updates of major dependencies sequentially
          • Jean-Marie: do Java then Python, Ice 3.6 still a problem
            • Ice 3.6 unlikely to make it into m1
          • Roger: difficult to combine CentOS 6, Python 2.7, Ice 3.5
            • Simon: test on Windows?
            • Roger: requires maintaining custom builds, using CentOS 7 would be easier
            • Simon: drop Ice 3.5?
            • Jason: it is acceptable to have problematic cells in the testing matrix
        4. Confirm requirement of GCC 4.6 for OMERO 5.2.

          • Already listed in our version requirements and allows moving forward with C++11.
          • Requires CentOS 6 C++ developers to use RedHat-supplied SCL devtoolset-3 package.
            • Chris: from Glencoe side on Linux, CentOS 6 is still what people use; seems fine though
        5. Confirm requirement of Python 2.7 for OMERO 5.2.

          • Already listed in our version requirements and recently requested by Ola.
          • May force a requirement for Ice 3.6 on CentOS6 since Ice 3.5 is built against 2.6. Our options are to build our own Ice 3.5 against SCL Python 2.7 (not recommended) or to recommend Ice 3.6 with the IcePy provided via pip, which will be supported by ZeroC. CentOS 6 needs Python 2.7 from SCL
          • Simon: need to know if C7 infrastructure (required for Python 2.7) is needed for m1
            • Jean-Marie: resources can be dedicated
            • Jason: good to plan stepwise timeline and understand how much resources is needed when
            • Jean-Marie: will have a planning/deciding call tomorrow
        6. Confirm removal of Ice 3.4 support for OMERO 5.2.

          • Already listed in our version requirements
          • Given the SSL/TLS cypher changes in Ice 3.6 and the size of the test matrix, it would also be helpful to remove this to reduce the inter-version operability and testing burden
          • Sebastien B: it is largely already done
          • Roger: will open a PR for remaining cleanup
        7. Split supported platforms into two tiers. It's clear that to make some of the changes we planned that supporting older CentOS releases will require some compromises--additional packages and setup by the admin. They can still remain "supported", but with some caveats.

      2. Spaces: decoupling, metadata

        • Sebastien B: decoupling space to end once Jean-Marie has brought Dominik’s gateway work into mainline
        • Simon: work on factoring away populate metadata + ROI scripts, requires further discussions
    2. Bio-Formats 5.1.4 - 14:54 UK

      1. Jean-Marie: had first Bio-Formats stand-up (1400UK daily), cleaned up and closed 5.1.3 board
        • Jason: cards to come for 5.1.4 board?
        • Melissa: plenty of cards there
    3. Glencoe Update (Chris) - 14:56 UK

      • prepping for CentOS 7, OMERO 5.1 upgrade
      • lots of populate metadata users
    4. Repositories - 14:58 UK

      1. Virtual Microscope/teaching (Ola)
        • deployed 5.1.3 on staging, Petr is testing, probably good for upgrade soon
      2. IDR Project
        • Jason: Eleanor has eight or nine screens on dogfish, so reviewing in planning work for coming year
        • Jason: planning fall visit to Dundee for Cambridge-area IDR people
    5. FLIMfit/OPT (Ian) - 15:00 UK

      • working on bugs, including Beatson Mac/PC issue
    6. Other consortium timelines (5 minutes) - 15:01 UK

      • Mark H: hackathon last week went well, work toward arbitrary n-dimensional chunks instead of being so plane-centric
        • also Curtis worked on testing scripts
      • Mark H: coming work on papers, conferences
      • Jason: URL for hackathon stuff?
        • Mark H: will send one out
  3. AOB (5 mins max - technical discussions should be highlighted to relevant people and rescheduled) - 15:03 UK

    1. Outreach cf. Jason's e-mail.

      • Jason: need to adjust standard presentations, many practice talks coming
      • Jason: important to schedule practice talks sufficiently in advance
    2. Thursday morning stats meeting.

      • June: calendar now updated
  4. Update on robot testing by Ola - 15:04 UK

    1. Balaji next week (Icy)
    2. Problems arising largely from Selenium, probably time to evaluate alternatives.
      • issues with both Firefox and Chrome
      • Chris: issues not unique to us, how does the rest of the functional testing community deal with this? The web community help each other and there are many Selenium users.
    3. Intermittently failing tests could largely be rewritten to improve, but not guarantee, their accuracy.
      • Ola: actually no clear path toward improvements
      • Jean-Marie: could use “broken” tag and separate CI job for problem tests
    4. Chris: could focus testing on what the framework/driver is better for
    5. Roger: mock some of the server-side to make client tests more reliable?
    6. Simon: at least run on single-threaded development server?
      • Chris: not as realistic
    7. Jason: could look at outsourcing environment actually running the tests
    8. Will: previous effort is actually now starting to pay off
    9. Ola: Don’t block merging PRs that adjust tests if failures are not from those tests.

16:04 UK

Document Actions