Greg Roodt

``` There are (at least) 5 different distributions of PyTorch for different accelerators (cuda 11.8, cuda 12.6, cuda 12.8, rocm 6.3, and cpu). Unfortunately, environment markers can't represent these conditions,...

> Yeah, that's actually the hack I'm seeing right now. A custom repo rule preprocesses a requirements.txt, adds torch@/some/local/path.whl, and then that is what gets passed to pip_parse. Sort of...

Oh nice! I've got a POC of a basic `uv` integration of option 4. The logic is not complicated and it's much simpler if we only need to support blzmod....

I think we will end up with various different kinds of rules and actions tbh. Probably some will be build rules, some will be repo rules, some might be macros...

> * Set the `hub_name` default value to `pypi` (it should not be `pip` because we are not using `pip` here and `pypi` is a more generic term). +1 For...

Yes, I should clarify as well. I'm only firmly +1 on the hub_name defaults. For the rest, I'm -1 Im not convinced on any lock file merging for similar reasons...

Yes, rules_rust has something similar for crates in cargo. You can load crates from a cargo lockfile, but you can also load them from specs in Starlark: https://github.com/bazelbuild/rules_rust/blob/b5ecaea4b5996912693135ea14447d22271e9cd5/examples/bzlmod/proto/MODULE.bazel#L77 ``` crate...

I made a proposal a while ago, but nothing has really progressed: https://github.com/bazelbuild/proposals/pull/275 I'm supportive of the idea, I'm just concerned about teams having to bring additional toolchains for compiling...

> The problem is that it requires Python **twice** There are sneaky things that can be done with the shebang so that it uses the hermetic interpreter, but I'll need...

I think they're quite different if I read this issue. I think this issue title is misleading, but you can confirm what you mean. 😊 1103 is for package publishers....