Ruff failing to start when used directly in notebook

[AWH1122]

When ruff is being used exclusively in a notebook it fails to start. Here is the log output:

2025-01-26 19:54:52.014 [info] Name: Ruff
2025-01-26 19:54:52.014 [info] Module: ruff
2025-01-26 19:54:52.014 [info] Python extension loading
2025-01-26 19:54:52.014 [info] Waiting for interpreter from python extension.
2025-01-26 19:54:52.088 [info] Python extension loaded
2025-01-26 19:54:52.111 [error] Python interpreter missing:
[Option 1] Select Python interpreter using the ms-python.python.
[Option 2] Set an interpreter using "ruff.interpreter" setting.
Please use Python 3.7 or greater.

If I create a .py file, select and interpreter, then restart the language server it works.

Steps to Reproduce:

1. Open jupyter notebook
2. Try to format document and notice ruff is not working

[dhruvmanila]

Thanks for the report.

Unfortunately, I'm not sure if this is a bug in Ruff extension because we use the Python extension to query the currently selected Python interpreter (refer to the screenshot, in the toolbar it says "Select interpreter" which is coming from the Python extension as it's unable to determine it for the project). As mentioned in the logs, one of the ways to resolve this would to use the ruff.interpreter setting. We're also looking into improving the UX for this part especially for the native server because we could skip this step if the extension is unable to get a Python executable (#598).

[AWH1122]

Doesn't specifying the path mean that this setting must be configured at the workspace level for Python projects that use different versions?

[MichaReiser]

@dhruvmanila do you know the answer to this question?

[dhruvmanila]

Doesn't specifying the path mean that this setting must be configured at the workspace level for Python projects that use different versions?

Usually, VS Code should be able to find the Python interpreter from the virtual environment that's being used by the project but if, for example, you're not using a virtual environment then it might be possible that VS Code is unable to find an interpreter to use here.

But, to answer your question, I think you'd only need to configure it for the ones where the editor is unable to find an interpreter. As mentioned we do want to improve the experience here to avoid depending on this behavior. And, come to think of it, it seems that ruff.interpreter is scoped to "resource" which seems too specific and it should probably be "window" scoped instead.

[AWH1122]

So since I experience this issue only with notebooks, if I am relying on different versions of python than what is currently specified in ruff.interpreter as It stands, I will have to change the interpreter at the workspace level?

It is strange because as soon as I open a .py file there are no such issues and everything works as expected. Its like ruff 'can't see' the selected kernel in the notebook.

[dhruvmanila]

So since I experience this issue only with notebooks, if I am relying on different versions of python than what is currently specified in ruff.interpreter as It stands, I will have to change the interpreter at the workspace level?

Just to clarify, do you want to use different Python interpreters for notebooks and Python files that are in the same workspace? Or, is it that you've configured ruff.interpreter globally and want to use a different version for a specific workspace that contains the notebooks?

It is strange because as soon as I open a .py file there are no such issues and everything works as expected. Its like ruff 'can't see' the selected kernel in the notebook.

Oh yeah, I don't think we look at the kernel information for the notebook in VS Code.

[AWH1122]

Just to clarify, do you want to use different Python interpreters for notebooks and Python files that are in the same workspace? Or, is it that you've configured ruff.interpreter globally and want to use a different version for a specific workspace that contains the notebooks?

It is the latter.

[dhruvmanila]

Got it, thanks. I'd recommend for now to configure ruff.interpreter for the workspace. In the future, we want to remove the dependency on the Python interpreter or at least make it optional.