chore: Support uv

[jstlaurent]

Changelogs

• Synch the dependencies in pyproject.toml with those in env.yml
• Add a some commands to run when using uv for local development

---

Checklist:

• Was this PR discussed in an issue? It is recommended to first discuss a new feature into a GitHub issue before opening a PR.
• Add tests to cover the fixed bug(s) or the newly introduced feature(s) (if appropriate).
• Update the API documentation if a new function is added, or an existing one is deleted.
• Write concise and explanatory changelogs above.
• If possible, assign one of the following labels to the PR: feature, fix or test (or ask a maintainer to do it for you).

---

discussion related to that PR

[cwognum]

Looks great!

Do you know if there's an automated way to keep things in sync? I assume not and I expect it's not that big an issue because we won't frequently update it, but just out of curiosity.

Finally, if you're curious, I would also recommend to give pixi a look. From the creators of mamba. It's built on top of Conda (and thus cross-platform and multi-language), fast (written in Rust) and feels like a more complete package manager (e.g. it has lock files, task manager). pixi actually wants to integrate PyPi and Conda workflows and has written its own PyPi resolver called rip. It seems though they are now switching to uv instead! See https://prefix.dev/blog/uv_in_pixi

[cwognum]

The failing test cases seem to be an upstream issue. Maybe we need to set a limit on pandas, datamol or rdkit?

[jstlaurent]

The failing test cases seem to be an upstream issue. Maybe we need to set a limit on pandas, datamol or rdkit?

I put an upper limit on pandas. It looks like rdkit tries to patch a function that's moved around in 2.2.0.

[jstlaurent]

Do you know if there's an automated way to keep things in sync? I assume not and I expect it's not that big an issue because we won't frequently update it, but just out of curiosity.

I don't believe so, unfortunately. We'd need to set something up ourselves in CI, if we wanted that.

Finally, if you're curious, I would also recommend to give pixi a look. From the creators of mamba. It's built on top of Conda (and thus cross-platform and multi-language), fast (written in Rust) and feels like a more complete package manager (e.g. it has lock files, task manager). pixi actually wants to integrate PyPi and Conda workflows and has written its own PyPi resolver called rip. It seems though they are now switching to uv instead! See https://prefix.dev/blog/uv_in_pixi

Hadrien had mentioned it. I had a look and it seems like the goals are pretty much the same as with uv (becoming Cargo for Python). I personally prefer uv's leveraging the Python standards through pyproject.toml, rather than using a custom format, like pixi's. It doesn't force potential users to use one specific tool.

[cwognum]

I agree! That's why I haven't switched, but they have some PRs on importing / exporting requirements to other formats.