Closed jvdd closed 3 months ago
We're aware of this—being a package that unifies a large and diverse combination of code derived from a range of different sources, it is challenging to keep code updated in this way. It's on our radar, and we welcome pull requests! As I understand, (and @jmoo2880 will correct me if I'm wrong) most of the SPIs do run on newer python versions, as does the main calculation infrastructure.
This is certainly something that is on our radar. As @benfulcher mentioned, pyspi draws from a range of dependencies to compute various SPIs. The downside is the potential for clashes and incompatibilities between these dependencies.
Specifying Python <3.10,>=3.8 offered the best compromise in terms of compatibility between the various dependencies (allowing a vast toolkit of SPIs) while also ensuring reproducible results across different systems. Naively bumping the dependencies could lead to inconsistencies in the computed SPIs, which is something we want to avoid. I touch on some of the difficulties with updating the dependencies in this thread: https://github.com/DynamicsAndNeuralSystems/pyspi/issues/60. Essentially, it requires careful testing and validation to ensure that the results remain consistent and reproducible.
@jvdd, I've managed to sort out the dependency issues described above and have released a new pyspi version 1.1.0
that brings support to all python >=3.8 (incl. 3.12). Let me know how it goes for you :).
Really appreciate the effort @jmoo2880!! Can't wait to toy around with pyspi :)
Currently
pyspi
requires Python <3.10,>=3.8. It would be awesome to see support for more recent Python versionsGiven that Python 3.12 is now the default installation in the latest Ubuntu LTS release underscores the need to expand the range of supported Python versions.![image](https://github.com/DynamicsAndNeuralSystems/pyspi/assets/18898740/de528ede-79b4-4e93-9066-79a1ae471328)