Current efforts - May 2025

[pgunn]

Recently I've been working on tackling topics that were blocking us from upgrading numpy and scipy and other dependencies; the most recent changes that landed (or will land soon) in dev tackle this.

Meanwhile we're still working on the tensorflow to pytorch port; when we're off tensorflow entirely, the release process will be a lot easier (Google doesn't seem interested in maintaining tensorflow as much as it has in the past, with Windows support reducing dramatically). After we're off tensorflow we'll also be ready to:

• Start splitting Caiman's visualisation components off from its core code, making its use by other packages simpler and pull in less stuff
• Move to a more per-run path system under caiman_data/ - we've been moving in this direction for awhile, and the hope is to provide relative isolation between simultaneous caiman runs, a more standard output process, and to make provenance of generated data better

Both of these will take us awhile.