Matt

@rickeylev pointed out that the toolchain itself is relevant here as well, so [here's](https://source.chromium.org/chromiumos/chromiumos/codesearch/+/main:src/bazel/python/toolchains/BUILD.bazel;drc=b01ff86d5215bd1ac2e0ad7cf3345b5a9fa9c76f) the link to our toolchain. We wrap the interpreter in a script that ensures that the...

> Those indeed should not be using workspace name as import prefixes. That's never been a documented or recommended approach. IIRC (I last looked at this around a year ago),...

In that example, `libs.my_lib` is defined in `//libs/my_lib` though. What we're asking is how to import a file in `@foo//libs/my_lib`. Before repo mapping, I believe that `import foo.libs.my_lib` would work,...

I'll go ahead and do that, but FWIW, I think that workspace name as prefixes is the correct way to go, and that if we fix that, we get this...

> Yeah, using the Bazel repo name as the Python top-level import name was always a bad idea. Even prior to bzlmod, because repos names were decided by the end...

I don't have any intention to add it myself, but If anyone does create a PR, I'd definitely like to review it. Simplest solution for now would probably to be...

In order to solve that issue, the only way to do it is to use the hub-and-spoke model, like I use in #1910. That also means you don't need to...

@scentini What do I need to do to get this reviewed?

I fixed the clippy errors that were causing the failures, but there still appears to be two failures which I believe are unrelated to my code.

Yeah, I worked around by renaming the file, running jj status, then renaming it back