---
tags: NGFF
---

# IDR-NGFF 2020-W48

## 2020-11-26
- Josh:
- social media on ngff repo
- then there's the question of v0.2 have a few days to figure that out
- Will:
- labels CLI PR to look at.
- playing with napari to stitch a plate
- did get it working locally (2D) with zoom-in
- tricky to get it working with s3.
- Seb:
- HCS is good for review
- one PR open for `acquisitions: []` fix
- regenerating data today, complete list on idr-zarr1-dev
- will try the same with bioformats2raw to compare
- David
- S3 rabbit hole ... Works for David but not for Josh.
- qupath doesn't look to handle multi-filesets well
- Petr:
- outreach in Goteborg, 8:10 UK
- miss standup
- need helper (also to miss standup)
- 2 hours with 20min break
- ends 1030 UK time.
- not J-m
- need a buffer since there's been

## 2020-11-25
- Josh:
- NGFF & bikeshed (from standup). Any comments?
- Merge all PRs into 'latest'.
- David:
- ZarrReader: weird dependency issue when bundled into a single jar with S3. Fine in fiji but not CLI. Just ome-commons causes the issue.
- Josh: may be ok to build an unbundled jar
- David: see uncommented section
- Testing in QuPath...
- Petr: preparing workshop feed
- How granular? Single I2K entry
- Seb: switching to GHA for quick review
- Seb
- working on latest HCS update (bikeshed experience)
- cf. bioformats2raw etc. from Melissa
- naming
- Latest == in-progress
- 0.1 == last release
- Will: are we going to 0.2?
- Josh: 0.2-dev? 0.2-rc1? Requires copying latest/index.bs to NEWDIR/index.bs
- Will: may depend on how much we show HCS. Not sure how much people are interested. People aren't likely to start working with many images at one time.
- Seb: HCS brings us the concept of "batch", first example of a spec that's a collection of images. Eventually, everyone wants that.
- Josh: esp. with the label overlay
- Petr: if you show just one image, then the questions come: "where are the others?", "what about looping?"
- Will/I2K:
- If we're going to show HCS, then we should have things merged.
- Agreed, to spec latest but actually 0.2 (or similar) then merge PRs and tag.
- FYI: https://github.com/ome/omero-guide-python/pull/27 more details on layout
- Finishing up PRs.
- Notebooks for showing analysis? Not really. More how to get a numpy array.
- Will: Notebooks local?
- Petr: mybinder requires caution (trick was: leave zoom, restart page, ... 5 min of debugging)
- Simon: 5 minutes is probably the critical bit.
- Petr: repo2docker doesn't work on Windows
- Simon: load on mybinder before the workshop
- Petr: but have redundancy locally in case it fails
- Will: was using vizarr example with `pip install` and `jupyter notebook` (*not* repo2docker)
- Petr: conda -> pip install repo2docker works. (Once it was necessary to install from master)
- Josh/I2K
- email with instructions:
- image.sc meeting
- image.sc HCS (if we get to 0.2)

## 2020-11-24
- Josh:
- biorxiv paper on image.sc (N5 plugin). N5-N5 format (Combining bdv.xml + n5). Janelia COSEM
- At minimum, helps with spatial context.
- Fixing Python 3.9, then back to PR testing
- Also, PPTXing
- Will:
- starting to think about what to show for i2k
- what can be hands-on
- actively code?
- mixture of people: python, java, javascript, non-coders
- Will: was imaginging zarrlite.js. play with parameters, render a plane. likely the simplest implementation.. likely the simplest implementation.. likely the simplest implementation.
- instructions for installing
- Petr: depends on who is coming
- Will: imagine _most_ want to learn _something_ (even if just imagej macros)
- Petr: don't want to give 18 the feeling they aren't welcome
- Josh: breakout rooms? Petr: available?
- what do we want them to understand
- PRs
- vizarr
- Will: merge conflict (various dependencies)
- Josh: want to move code to S3. would like to have it merged.
- switching to https://hackmd.io/qmr2e48-QaqFlXNuMc-Fmw

## 2020-11-23
- JRS:
- Will's links work better on Campus
- Demoable. Make video at school?
- Josh: warn everyone in terms of bandwidth boundaries.
- Small list of fixes for usability
- e.g. if you change the rendering at one level, the rendering isn't passed to the next level. i.e. be prepared to hear about them.
- Will/I2K
- choosing people. Several at level one? Seb: yes (20 is crazy)
- how many? 10-20? 130 applicants. 40 at 1st.
- Petr: prepare exercises or no hands-on (for this time length)
- Will: could update local code to read pixels from ome-zarr
- Simon: what do we want?
- Josh: hope to get 2-3 people to engage after the workshop
- Will: take a large number of people, but nothing interactive
- David
- hopefully solving multiscale issue today
- Petr
- who was the audience for Friday? (seemed sub-optimal)
- Josh: think for visualization it's clear. For analysis it's difficult.
- Petr: perhaps only for batch analysis, but ilastik is not an easy beast.