fix: Drop python 3.8-3.9, upgrade deps, use ruff format

[max-sixty]

No description provided.

[max-sixty]

Bumped this all the way to 3.10 since otherwise requires careful management of ipython dependency

Colab is on 3.10, which is some standard

[eitsupi]

Thanks for working on this.

Bumped this all the way to 3.10 since otherwise requires careful management of ipython dependency

Colab is on 3.10, which is some standard

Hmmm, I don't know if it's worth dropping Python 3.9 just for that.

Isn't Colab based on 3.10 because python3 on Ubuntu 22.04 is 3.10?
https://launchpad.net/ubuntu/jammy/+package/python3

That has nothing to do with the release cycle of Python itself (since Ubuntu LTS is released every two years)

[max-sixty]

I don't think it's related to Ubuntu: https://colab.google/articles/py3.10

[eitsupi]

Thanks for the context!

In any case, I don't know if there's any justification for dropping Python 3.9, given that duckdb, pyarrow and polars support 3.9.

[max-sixty]

I agree it's not ideal. But pandas etc have dropped 3.9, and it seems like we're going to need to do some dependency surgery to make ipython work with it — I'm mostly thinking whether that's worthwhile...

[eitsupi]

I don't understand why the pandas version is relevant here.
pandas does not seem a dependency of IPython.

[max-sixty]

Yes, I just mean that pandas is a big library that others follow — it's not like every other library support 3.9 except for us. Does that make more sense?

[max-sixty]

I think unless someone wants to do the work to make it py-3.8 compatible, it's a net benefit to merge this...

[eitsupi]

I think unless someone wants to do the work to make it py-3.8 compatible, it's a net benefit to merge this...

I don't understand why you treat Python 3.8 and 3.9 the same, Python 3.9 is still supported right?

[eitsupi]

I don't understand why the pandas version is relevant here.
pandas does not seem a dependency of IPython.

[eitsupi]

The badge shown in the README needs to be updated.

[max-sixty]

I don't understand why you treat Python 3.8 and 3.9 the same, Python 3.9 is still supported right?

3.8 & 3.9 are both difficult to get work; they require making custom IPython dependency specifications. The easiest way to see is to try and build this while supporting 3.9...

[eitsupi]

Given that upstream jupysql supports Python 3.9, I am not sure if your argument makes sense. In my opinion this is simply a limitation due to current CI and poetry, not a limitation of the Python package itself.

[max-sixty]

I agree it's surprising. And I'm sure it's possible to support 3.8. It's just a tradeoff of whether it's worth the investment.

Big fan of uv, I use it myself. FWIW it required more rather than less work to get xarray working with it (but once it does, it's reliable and really fast)