DetachHead

> (Would love to hear other solutions) the solution i'm using for now is [pyprojectx](https://pyprojectx.github.io/). i mentioned it in the OP but i didn't really make it clear that it...

> You get it to team members by having them install whatever version of uv they like (that's new enough to have this feature) via the official install script, pipx,...

> Except they don't need a version of Python installed at all. Installing uv doesn't require Python as a pre-requisite when using the official installer script i'm not a fan...

i'm not very familiar with jupyter notebooks. is this what you mean by %sql cells?  pylance seems to have the same issue: https://github.com/microsoft/pylance-release/issues/4969 interestingly, this other syntax with `%%sql`...

could you provide a bit more info please? what exactly is the difference between `%` and `%%`? is databricks a separate thing to notebooks? a link to some relevant docs...

does `native-tls` work for your use case? https://github.com/astral-sh/uv/issues/6572#issuecomment-2594166728

[execution environments](https://docs.basedpyright.com/#/configuration?id=sample-config-file) allow you to use different diagnostic settings on different parts of your project. however i'm not quite sure what you mean by applying different rules to third party...

interesting idea. i'm not famiar with pandas though so i'm curious to see some examples of why it can't be fully typed

i guess `reportUnknown*` vs `reportAny` covers that

> The issue is that these "problems" usually don't even mean anything, you often have to suppress them, and then suppress real problems with your code etc. > > TL;DR...