Nix would be another option. I find it odd so many tools like this are so Python specific, when a lot of the complexity is dealing with non-Python dependencies (glibc versions, and all the different C shared libs).

I think there's room for both. I want speed and correctness within my little Python ecosystem, so I reach for a tool like Rye to help with the package management. Then to solve the system level issue maybe I reach for a simple direnv implementation to install the system level dependencies needed for the project.

You could of course lean all the way in and manage and build your Python dependencies using Nix as well, but that is... introducing a lot more complexity and slowness at this time.

I think it comes down to the frequency at which we iterate on things and the level of expertise required to deal with them. The system level dependencies might change much more rarely, and require more expertise to work with tooling, but if I can hand off a clean system environment to a developer who doesn't know about or need to care about Nix at all and then they are able to quickly iterate inside the Python ecosystem I get the best of both worlds - I solve the non-Python dependency problem myself and the end user (the developer) has a fast, smaller scoped, way less inscrutable problem to deal with when they are working within the Rye ecosystem.