2nd December, 2020

Attending: Jamie Sherman, Ryan Abernathy, Josh Moore, Matthias Bussonnier, Ryan Williams, Jim Pivarski, John Kirkham

• Josh workshop:

  • API maybe the partial read; instead of having multipld copies;

    can each chunk be compressed in a better way (progressive JPEG) how would that look like to a user.

  • Could be implemented as a codec

  • Ryan: premature optimization. Zarr is focused on scale out; this

    seems low-level optimization.

  • Josh did selling of rechunk, peoples are quite happy.

• Zarr 2.6.1 released; first version to be 3.6+ only. Screwup with 2.6.0 which was actually published as 0.0.0 due to build isolation and setuptools_scm, waiting for the conda-forge bots.

• CZI EOSS next wednesday (1900 UTC next Wednesday)

  • Matthias rehearse :

    https://photos.app.goo.gl/H91qwgNDwfcUCE2K6

  • https://docs.google.com/presentation/d/1O4_J8f6dGeV9I7NntPqPtSRDPeJLSjNOfQstI4BB-_I/edit?usp=sharing

  • Sending links to slides & video

  • David video:

    https://drive.google.com/file/d/1fInOf_9JiNCTyMX5BYsKQS0vQpFTzuau/view?usp=sharing

• Changing fill value (Matthias for Ryan)

  • easier way to do it.

  • expose as property on the array, print warning on setting

  • John: expose is good. warning should include previous.

  • Josh: will it change which chunks will be written to disk? No.

  • Matthias: assume the use case is changing after creation before

    writing

  • https://github.com/constantinpape/zarr_implementations

  • https://github.com/ome/ome-zarr-py

  • https://github.com/ome/omero-cli-zarr