Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

@f0rmiga would you be willing/able to have a look? From the associated issue, it sounds like you would be better qualified to judge.

At a glance, I would have expected cc_import to Just Work, since it's doc indicates it's for prebuilt shared libraries, which would imply either (a) something downstream of it is incorrect, or (b) something is missing in our cc_import() call. But I don't know enough about the difference between cc_import and cc_library (and the lower-level linker stuff mentioned in the issue) to say with any certainty.

I just came up with another solution for myself that completely avoids having libpython as a dependency for my pyo3 target. I should have thought of this ahead of time, as this is important for manylinux compliance.

I enabled the "extension-module" and "abi3-py38" features for pyo3 and now I don't need libpython as a dependency

I just came up with another solution for myself that completely avoids having libpython as a dependency for my pyo3 target. I should have thought of this ahead of time, as this is important for manylinux compliance.

Great, does that mean changing it to cc_library is unnecessary now? If so, if you want to pivot this PR to adding a comment, e.g. # NOTE: if you're depending this from rust, then ..., we can finish review. It sounds like a good hint to give future people trying to do rust+python interop

I still think the approach in the PR is correct. In practical terms, the difference between cc_library and cc_import is that cc_import only takes 1 file, so in the case of Linux that has 2 files (the real shared object and the symlink to it), there's no way to use cc_import. For one case I worked on, the cc_import was fine because we were using RBE, and since RBE converts the symlinks into the realpath contents, it worked fine.