New release: Brian 2.5.3

Announcements
1

We recently made not just one but two[1] new Brian releases :tada: !

Unfortunately, a number of major enhancements are still in the work-in-progress state and did not make it into this release. For users, the releases therefore mostly fix incompatiblity issues with the latest versions of dependencies such as numpy, but also a fair number of other minor bugs and issues. Contributors (or future contributors) might want to have a fresh look at our code base and marvel at its newly found black-formatted consistency :eyes: (see Coding conventions — Brian 2 2.5.3 documentation for more details).

As always, find a detailed list of changes in the Release notes — Brian 2 2.5.3 documentation.

Thanks to everyone who contributed to this release, and stay tuned for the upcoming release, hopefully packed with new features!

  1. Admittedly, the second one was only necessary because the documentation and conda packages were no longer building correctly – thanks to @mpagkalos for making us aware of the documentation issue! ↩︎

4 Likes
2

We noticed that this release is limited to python>3.9. Is that a hard requirement for Brian now? Python 3.8+ are still supported (3.7 just went EOL last week, so will still be floating around for a bit). So, it’ll be good to maybe support py3.7+ for a bit?

For things like the OMV framework, we try to support a few versions of Python because folks don’t always update their Python versions as soon as they go EOL.

3

We decided to follow numpy, matplotlib, etc. (NEP 29) – this makes it much easier for us maintenance-wise, and we can pick up newer Python features with only a ~ 3 year delay (e.g. now we could use dict unions and the graphlib module).

1 Like
4

For anyone using the new version and interested in GPU-acceleration, please note that brian2cuda and brian2genn now run with the latest Brian version as well: New releases: Brian2GeNN 1.7.0 and Brian2CUDA 1.0a3