Sebastian Berg

The interesting line is (snipping and breaking it for clarity). ``` Original error was: dlopen(.../python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so, 0x0002): tried: '.../python3.12/site-packages/numpy/_core/_multiarray_umath.cpython-312-darwin.so' (mach-o file, but is an incompatible architecture (have 'arm64', need 'x86_64')), ```...

@mtsokol do you happen to remember if there were changes related to `copy=` in 2.1?! It seems to me that this is a an actual regression from 2.0, in that...

Sorry, nevermind... This seems to be a pandas bug. NumPy passes `copy=True` here, and pandas must not ignore that (but it does ignore it). @lil-blue-curve could you open a pandas...

Thanks for looking into this @nitinvunnam. Could you move this fix down one level? And maybe be up for a bit more maintenance here? The `_routines_manipulation.pyx` file has a `_normalize_axis_tuple`,...

After discussing a bit and trying to understand it better (correct me if I'm wrong)... I don't think we should discuss this as a proposal to actually change the standard...

@r-devulap probably knows if there is any recent change (not sure). In general, I am not sure this looks too bad. Different math libraries behave differently with corner cases and...

The original blocklisting PR, mentions for example: ``` np.tanh(1000+0j) gives nan+nan_j instead of 1.0+0j. np.tan(0+1000j) gives nan+nan_j instead of 1j. ``` which is clearly worse then inconsistent results for NaN-values....

It looks like at least the old intel compiler doesn't like `_Complex` in C++ (on windows?). My suspicion would be that needs to be tweaked, half of the branching necessary...

This has a bit convoluted reason: * `cupy.cuda.runtime` exists as a module to re-export everything in `cuda_backends...runtime`. * Re-export with `from ... import *` omits private symbols (which is OK)...

Thanks for the report. I think this is essentially a duplicate of https://github.com/cupy/cupy/issues/3513. Cupy sparse matrices are 32bit only right now (which should be changed), while scipy will switch between...