Added support for numpy>=2

[fabiocat93]

This addresses issue #9

All unit tests pass. Compatibility with numpy>=1.21 is maintained

[iver56]

Did you try to run the CI with numpy v2? It would build for all targets (various OS, python versions, CPU types etc) and run tests for each of them

[fabiocat93]

I haven't let me try

[fabiocat93]

[WIP] fyi: CI fails also with on the main branch which i haven't touched

[fabiocat93]

Did you try to run the CI with numpy v2? It would build for all targets (various OS, python versions, CPU types etc) and run tests for each of them

I ran the CI with numpy v2, and all CI checks passed. The builts succeeded for x86_64 and ARM64 on ubuntu-latest and macos-latest. I skipped Windows for now, as it fails in the main branch as well. I believe this is ready to be merged @iver56

[iver56]

Sorry, but removing Windows support is not a viable way forward right now. I'll have a look at what fails in the main branch right now, and see what needs fixing.

[iver56]

Looks like this is the issue we're running into: python-cffi/cffi#118

[iver56]

PyPy includes its own version of cffi, and I don't know when they will create a new release with the patched version of cffi. What I can do for now is to specifically disable pypy3.9 builds for Windows. Python 3.9 is nearing end-of-life anyway, so this seems acceptable.

[fabiocat93]

PyPy includes its own version of cffi, and I don't know when they will create a new release with the patched version of cffi. What I can do for now is to specifically disable pypy3.9 builds for Windows. Python 3.9 is nearing end-of-life anyway, so this seems acceptable.

sounds reasonable to me. I assume you don't need help with that, but let me know if you face any more issue I can give a hand with

[fabiocat93]

@iver56: Quick update: I've continued work on my branch (numpy2 branch) and updated the CI configuration to include Windows builds by removing Python 3.9 (since it's nearing end-of-life and was not building well). All tests now pass as you can see here: CI run results. One interesting note: the pp311-win_amd64 target was failing tests, so I've excluded it to keep the CI stable. If you’re on board with these adjustments, I can open a new PR with these changes. Let me know your thoughts or if there’s anything else you'd like to address.

Thanks!

[iver56]

Looks like we arrived at a similar conclusion - I have actually already commited similar changes: https://github.com/nomonosound/numpy-rms/commits/main/
Sorry for not keeping you up to date. Release coming soon.