project mew: A microbenchmarking CLI for Python

[nicholasjng]

This isn't a bug report, but rather a "Show and Tell" type discussion.

I spent a few hours in the past two weeks and came up with a new set of Python bindings for Google Benchmark. The repo lives at https://github.com/nicholasjng/mew, and is open-source.

Aside from feedback on the UX, which I always appreciate, there are (in my opinion) some parts that could be useful for GBM's own Python bindings, specifically the new PausedTimingScope RAII class binding as a context manager. Happy to hear your thoughts!

---

AI disclosure: I used AI to plan and implement parts of the project.

[dmah42]

it looks very clean. is there a passive request that we stop developing the bindings in this project and forward folks to mew for python use-cases? or do you rely on the python bindings in this repo?

[nicholasjng]

@dmah42 not from me, I'm not sure who depends on the GBM bindings.

I'm happy to upstream any parts you find useful, just let me know.

[nicholasjng]

Quick update, the project is now available on PyPI as mew-bench. After some considerations, I decided to eliminate external dependencies entirely, so the package+CLI is stdlib-only on top of the binding core, and comes out to a ~200KB wheel download per platform.