Nathan Goldbaum

Thanks for all the comments! I hope the new text addresses the concerns and is clearer. I'd like to add sections on the critical section and PyMutex wrappers, assuming those...

Well, unfortunately, nothing is ever as easy as I'd like. The issue Ralf pointed out was caused by me unintentionally using the system pip. After making sure there's a pip...

Also worth noting that I can reproduce all these failures locally on Python 3.11 on Linux using UV's python install.

> Might be worth reporting the bugs to python-build-standalone and see if they get responses/fixes? Yup, reporting these issues is on my backlog now. > If so, this PR will...

Closing in favor of https://github.com/numpy/numpy/pull/27707

I agree this definitely makes sense in an external package. Starting with an external implementation that we could think about importing into numpy would also motivate the discussion about including...

See e.g. [this CI run](https://github.com/numpy/numpy/actions/runs/12953237446/job/36132442477?pr=28181) as well as [this PR from December](https://github.com/numpy/numpy/pull/28032#issuecomment-2552979977).

The main issue here isn't so much the implementation of changing the default fill value, or even coming up with better default fill values, it's whether or not changing this...

Hi, I don't think this interaction with dask has been reported. Can you make a short runnable example that demonstrates the issue?

ping @seberg in case you weren't aware of this interaction between dask, masked arrays, and the NEP 50 changes Maybe we should use the minimum or maximum int as the...