Nathan Goldbaum

> Although the documentation suggests arr += np.asarray(128) could be used instead, this doesn't always quite work. For example It does work if you specify the dtype though: ``` >>>...

I think all you need to do is set `__name__` manually for the ones that are actually reductions: ``` >>> print(pydoc.render_doc(np.ma.alltrue)) Python Library Documentation: method reduce in module numpy.ma.core reduce(target,...

@gangula-karthik that seems reasonable to me, if you can try sending in a PR that would be great. Also maybe a test to make sure that `__doc__` and `__name__` are...

It would be nice to get new asv benchmarks (see the `benchmarks` directory) for mgrid and ogrid performance. AFAICS we don't have any benchmarks for those yet. Generally we like...

Unfortunately `bool` isn't subclassable, so `np.bool` is its own separate thing. It's backed by 0 and 1 8-bit int values.

Closing, fixed by #26896. btw @eagunn if you put "closes #xxxxx" in your PR description, when the PR is merged the issue will close automatically.

I did some local testing on this PR. After rebasing on `main` to fix some noisy tests that were fixed by me last week, this does seem to fix the...

> It might well just work if those two lines were removed. Nope, it leads to a compiler crash: ``` Error compiling Cython file: ------------------------------------------------------------ ... # Iterator API added...

We released PyO3 0.22.6 which should hopefully fix the segfault. However, it turns out that pip's wheel cache may need to be purged to actually build a new wheel. After...

> 0.22.6 doesn't fix the segfault, but it at least causes installation to fail at install time in the right package without UNSAFE_PYO3_BUILD_FREE_THREADED, which I think is a big improvement...