3rd July, 2019

Time: 20:00 BST (15:00 EDT).

Joining instructions see above.

Attending: Alistair, Josh, RyanW, Ward, John, Martin

Standing items

Introductions: new joinees should feel welcome to introduce themselves and their interest in zarr/n5.

Items

  • Telecon frequency, length, format. (20:04)

    • General agreement on proposal. Revisit time slot, lower

      frequency, potentially shorten. Though may need more time if we’re meeting less frequently.

    • Also need a rotation for sending reminders, etc. Thoughts?

    • Ryan: would say keeping every two weeks is good, even if don’t

      have stuff to report. Once a month in the throes of a spec seems long.

    • Ward: could accommodate either schedule. Do like the momentum.

      Every two week with nothing to report helps to keep the momentum. Can go to less frequent later.

    • Martin: no one wants to take on more organization. Ad hoc

      discussion is important.

    • Josh: could also see more informal meetings (even down to 15

      minutes) and use it to set up longer meetings.

    • Alistair: can see more frequent, but would help massively if

      someone else is sending the reminders. Cf. w3c – 1.5 hours every couple of weeks, and then periodic F2F meetings.

    • Josh: and biweeklies are still reporting and deeper technical

      discussions?

    • Alistair: planning on focusing on the spec and python

      implementation, so would be good to have time to surface design issues.

    • Ward: reasonable.

    • Alistair:

      • doodle after scipy (new connections)
      • skip the 17th, likely to re-start the week after (@John to send)
  • CZI call (20:18)

    • Alistair: interest expressed, but concretely …

      • Also starts towards the end of the year

      • Rough roadmap: (1) finish v3 core spec, (2) implemented in active languages (C, Scala, Python, C++, Julia, ~Java (native, re-use scala, wrap C)), (3) fundamental extensions like codecs & storage layers and perhaps protocol extensions, (4) governance (sustainability / transparency / fairness)

      • Someone to work on spec & implementation

      • Cf. new use cases (MET office, sparse arrays, etc)

      • Josh: thinking of very similar steps as Alistair – (0) governance (1) spec (2) prototype (3) supported format – and that works in parallel with the zarr plan.

      • Alistair: happy to host someone

      • Matt (via Alistair): talk to people at scipy about identifying peoples

      • Ryan: wondering via spec v. implementation pitch. Could say both. Personally have some bandwidth to work on implementation. Could potentially host someone at Mount Sinai.

      • John: We could try to get affiliated with NumFOCUS. That’s a bit easier. Alistair and I could try to get this started at SciPy

      • Alistair: looking for ways to accelerate development, failing that we will try to get people together. “Here are the people who are looking to contribute.

      • Ryan: NumFOCUS, 250K (100K for spec; 100K for development work, 50k overhead)

      • Unidata? Ward: have put in several NSF grants (“cloud native data access”, explicitly stating zarr) Attractive.

      • Josh’s three pillars of Zarr excitement:

        • bio domain applications (imaging, single-cell, genomics,

          pangeo, etc.)

        • new functionality / cloud-native features / distributed

          processing

        • integration with existing stable NetCDF standardized

          layer / ecosystem

    • TODOs

      • RyanW: to write to NumFOCUS

      • Alistair/John: to follow up at scipy

      • Alistair to email to all interested parties

      • Josh: will send gdoc for bioimaging

      • [???] governance for NumFOCUS

        • Ward: hdfgroup has governance documents
      • Add to the blog!

  • How to break up the spec into a series of discussion topics,
    and what order to take them in. (20:xx)

  • Roadmap?

  • v3 spec latest draft

  • v3 design update

  • SciPy / PPTX discussions

    • Alistair: best exemplars? Thinking to give 3.

      • Malaria genomics
      • Single cell (from RyanW, Ambrose as a backup)
      • Pangeo (RyanA)