Evgeni Burovski

Yes, this is unfortunately a known issue with CuPy that it requires a context manager. NOt sure if there are plans to change that, maybe @leofang could weight in?

Potentially, yes. The key question ATM is what's the cupy team plan. If this is something they plan to fix, we can either wait or potentially help materializing for cupy...

This seems to have slipped under the radar. Could you remind me what are the motivating use cases? IIUC, one was float32->complex64, which is now IMO adequately handled by `1j*float32_array...

This is the int/uint discussion I was trying to remember: https://github.com/data-apis/array-api/issues/859#issuecomment-2502160704

Here is a quick session of downstream testing: I ran test suites for SciPy, scikit-learn, scikit-image, networkx and pandas. Short version: scipy has small issues which I'll take care of;...

> I am aware of the array api work, but this should not come at the expense of giving up the convenience of a dynamic language. Array API discussion is...

> I am assuming this is the reason but otherwise what's the reason to change it? The array API discussion linked above is a context where this issue originates, yes....

1. The failure in the scikit-image test suite (in the OP) is basically https://github.com/scikit-image/scikit-image/issues?q=is%3Aissue%20state%3Aopen%20eig --- there are several issues with eigenvalues becoming complex _sometimes_. The change will make them always...

The `scikit-image` solution is two-fold I believe. What is the problem. The failure is in this line [1]: ``` phi %= np.pi ``` with `phi` is a numpy scalar computed...

And for completeness, _the article_, Ref [1] in the scikit-image code, is https://autotrace.sourceforge.net/WSCG98.pdf