Bradley Dice

@leofang Happy to help review this packaging work! Please tag me on any PRs. 👍

Hi, what is the next step for this fix? I just ran into this while attempting to help conda-forge with the Python 3.13 migration for the `conda` package. This does...

The CUDA runtime docs show it here: https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html > cudaDevAttrMemoryPoolSupportedHandleTypes = 119 > Handle types supported with mempool based IPC To the best of my knowledge, this enum value has...

Thanks for the issue! Here are some notes on what is going on. Enabling statistics wraps the current MR in an adaptor that tracks statistics ([see implementation](https://github.com/rapidsai/rmm/blob/dcd37acbe58c73192f74bd6bb1ec5f1b9f1f2dd4/python/rmm/rmm/statistics.py#L67-L71)). Similarly, enabling spilling...

In theory we should define `__all__` for every public API in the namespace that we intend to expose it in. So maybe this is only needed for package `__init__.py` files,...

This introduces a bit of complexity that we may be able to avoid. RMM declares its minimum supported CUDA version is 11.4 (this has been true [since November 2023](https://github.com/rapidsai/rmm/commit/7f776eff99b65ffe7b41bc547ec5d0a5dde3cb61)). We...

@kinto0 Thanks! I had a chat with the Pyrefly team at PyCon. It sounded like most of the concerns were around the general experience with git pre-commit hooks but not...

Thanks @lolpack @yangdanny97! The `system` hook is a good starting point! As a follow-up, using a `system` hook makes it hard to constrain the version of pyrefly that is used...

Sorry for the delayed response. Here are some replies to the questions above. > From the mypy pre-commit hook, it takes in arguments for `args` and `additional-dependencies`. These are pretty...

That sounds like a good starting point! I'm not aware of any other blockers yet, because I haven't pursued Pyrefly any further after getting blocked by the lack of a...