provide Python builds for linux-aarch64-musl

[mikolysz]

linux-aarch64-musl is the platform one gets when using an Alpine-based Docker image and developing on Apple Silicon, which is a pretty common setup at this point. This set up currently (as of uv 0.4.1) doesn't support installing Python via uv, with the following error:

Searching for Python installations
error: No download found for request: cpython-any-linux-aarch64-musl

For other users encountering this issue, installing Python via app add python3 seems to work for now, although Alpine's packages tend to be fairly minimal, so more dependencies might be needed for larger projects.

[charliermarsh]

This is known but can't find an existing issue to link to -- will leave it up for now.

[charliermarsh]

Related to #6643.

[zanieb]

Yeah more context in

• Inspect system to determine if musl toolchains should be used #4242
• uv venv --python <version> fails when querying managed python interpreter on alpine #6392

[SamuelMarks]

What's the current progress here? - Need a hand?

[zanieb]

If you're interested in lending a hand it's definitely welcome! I haven't started working on this part of python-build-standalone yet.

Briefly, it looks like there were some previous discussions:

• Allow dynamic loading in musl builds while CPython itself is static python-build-standalone#86
• MUSL distributions lacking proper interpreter information python-build-standalone#260
• Drop support for Windows static builds python-build-standalone#221

astral-sh/python-build-standalone#86 seems like the best place to discuss implementation details, if necessary.

[SamuelMarks]

Ok, just took a look: https://github.com/astral-sh/python-build-standalone/releases/tag/20241219 doesn't have a build for my architecture + Linux + musl.

I'll get my other laptop back in a couple of days will try then

Happy New Year!

[mwarkentin]

I ran into this when attempting to use uv inside of an Alpine docker container (which is the first "derived image" mentioned on https://docs.astral.sh/uv/guides/integration/docker/#available-images.

My use case: working on a docker container that contains a grab bag of scripts that our team has built over time, and hoping to use uv to manage python versions and dependencies which may be different for each script, using the inline /// script definitions.

Will switch over to a Debian image for now instead.