MAINT, BUG: sync with upstream to unconditionaly disable use of aligned_alloc

[tylerjereddy]

• sync with upstream to deal with BUG: SciPy 1.13.0 not buildable on old macOS due to pocketfft code scipy#20300; as discussed there, our more conservative shim failed to fix the conda-forge issue for older MacOS, so let's turn this functionality off completely like NumPy did at BUG: update pocketfft to unconditionaly disable use of aligned_alloc numpy/numpy#26123

[tylerjereddy]

I'm going to self-merge, trivial sync to upstream needed for release.

[rgommers]

@tylerjereddy @lucascolley I think this is starting to go in the wrong direction. This repo is meant to be an exact mirror of upstream (EDIT: see scipy/scipy#19882), so we shouldn't see any diverging commit hashes.

I'm thinking of doing the following:

• push a new branch containing the diverging commits to preserve them
• bring the default cpp branch back in sync with the upstream cpp branch
• disabling pull requests and setting branch protection such that only regular pushes to the default branch are possible.

[tylerjereddy]

fair enough

[lucascolley]

I'm thinking of doing the following:

push a new branch containing the diverging commits to preserve them
bring the default cpp branch back in sync with the upstream cpp branch
disabling pull requests and setting branch protection such that only regular pushes to the default branch are possible.

Would you like to do this now? Let me know if I can help

[rgommers]

Would like to, but realistically not going to any time soon (too many other higher-prio things pending). If you'd want to do this, that would be great! As long as all the commits remain on a branch in the repo, this should be safe enough. Let me know if you want to go ahead and need more permissions.

[lucascolley]

Would like to, but realistically not going to any time soon (too many other higher-prio things pending). If you'd want to do this, that would be great! As long as all the commits remain on a branch in the repo, this should be safe enough. Let me know if you want to go ahead and need more permissions.

I'm happy to! And yes, I think I need more permissions.

[rgommers]

And yes, I think I need more permissions.

done (... full access to the project, including sensitive and destructive actions ...) - with great power and all that:)

[lucascolley]

@rgommers I've pushed a new branch commit-archive, but I am unable to force push to the cpp branch due to GH006: Protected branch update failed for refs/heads/cpp

[rgommers]

I'll have a look. Is https://github.com/lucascolley/pocketfft/tree/cpp up-to-date?

[lucascolley]

I think I've figured out how to disable the branch protection temporarily

[rgommers]

Good - yes you should be able to.

[lucascolley]

scipy:cpp now matches mreineck:cpp. I'm not sure how to disable pull requests, though

[rgommers]

Great, thanks! I don't think pull requests can be completely disabled, but I edited the branch protection rule to require 3 approvals, as a strong signal that PRs shouldn't be merged. Will open an issue to describe the state of things.