Developer meeting summary 25/05/2023

[alessandrofelder]

Retrospective for period Jan-May 2023

Went well:

• maintenance burden reduced thanks to numba refactor and maintenance work
• good progress and momentum, thanks largely to @dstansby
• memory issue solved
• next priorities clearer

Didn't go so well:

• existing tests haven't caught real-life data-related bugs
• priorities took a while to crystallise
• work management could improve (tech debt, context-switching/multi-tasking, asymmetric workloads)
• could have abandoned dask route sooner.
  We have plans around how to address the above, which is good!

Learnt:

• numba, dask, neuroscience, what brainglobe should look like to the user, current brainglobe structure
• most users like BG!

Still puzzling:

• best way to transfer knowledge
• confusing naming of existing package for historic reasons
• encouraging external contributors
• consistent developer experience in napari (we decided to consistently derive from QWidget instead of using magicgui).
• best way to maintain momentum and funding stream long-term
• how good is good enough

New release!

Cellfinder-core 0.4.0 has been released 🎉

Key priorities for immediate future

• Two new developers joining in mid-June, with @dstansby leaving. They will focus on Brainglobe (meta-) package and benchmarking/profiling respectively.
• Napari Atlas viewer widget
• Add tests for large scale data
• new docs website (we will follow what napari does)

Other possible things to explore in the future

• use napari-matplotlib to add some analysis/data exploration features might be cool!
• Contribute to napari group layer