SPEC 0 minimum supported python version

[d-v-b]

SPEC 0 recommends that we drop support for a Python version for 3 years after its initial release. That means dropping support for python 3.11 (released on October, 24 2022) soon.

Do we want to follow the SPEC 0 recommendation and drop support for python 3.11 soon? It's just a recommendation, so we have some leeway here.

And if we don't want to follow the SPEC 0 minimum python version recommendation, when would we drop python 3.11? 4 years after release? 5?

I would like to be guided by a signal from downstream projects. If literally everyone downstream has already dropped python 3.11 (big doubt), then we could drop 3.11 painlessly. If nobody is ready to drop python 3.11, then we incur the maximal pain if we drop it. We need to put a decision boundary at some point between these two extremes.

[maxrjones]

I think at least a few people thought that Zarr-Python follows SPEC 0 since it's listed as an endorser, based on the reactions to https://github.com/zarr-developers/governance/issues/11#issuecomment-2132184759

[jhamman]

I like SPEC 0 because it simplifies decision making and gives us a community reference to base our decisions on. Here's how I'd like us to think about the policy: we use the spec as a guide as to the "earliest drop date" and we drop support for old version when it is advantageous for us as a project. So we could drop Python 3.11 support as soon as Oct. 24 but we could keep it around for a bit if its not bothering us.

I would suggest we drop it when we add test support for Python 3.14 which could be quite soon I suppose.

[d-v-b]

FYI there's a proposal to amend SPEC 0 to support projects that want a slower drop cadence