MAINT: Adjust the codebase to the new np.array's copy keyword meaning

[mtsokol]

Hi!

This PR addresses changes planned for NumPy in numpy/numpy#25168 (new copy keyword for np.asarray and np.array).

np.array(..., copy=False) will now throw and exception if a copy is needed. To retain the same behavior np.asarray(...) can be used, so a copy is made only when needed.

I applied changes in a backward-compatible way, so that no NumPy version check & branching is needed.

[mroeschke]

If you have time could you investigate making test_array_interface compatible with older versions of numpy?

[mtsokol]

numpy/numpy#25168 got merged today - once a new nightly build is published, all tests should pass now.

[mtsokol]

NumPy nightly build disappeared (numpy/numpy#25907) but once it's back this PR should be ready.

[matthewfeickert]

@mtsokol Please retrigger the CI given numpy/numpy#25907 (comment).

[mtsokol]

Hi! I updated the PR with all required changes to be compatible with NumPy dev - now Numpy Dev CI job passes.

My question: There's a new copy keyword required in the __array__ protocol implementations. Would you like to keep this keyword unused (as dtype is sometimes ignored in the codebase) or do you want to actually make use of it internally where it makes sense?

[mroeschke]

Would you like to keep this keyword unused (as dtype is sometimes ignored in the codebase) or do you want to actually make use of it internally where it makes sense?

Yes we would probably want to make use of it, but that can be done in a follow up

[mroeschke]

Looks like there's some typing validations with the new addition of the keyword. Let me know if you'd like help with them: https://github.com/pandas-dev/pandas/actions/runs/8143484703/job/22255440584

[mtsokol]

Looks like there's some typing validations with the new addition of the keyword...

These errors look pretty tricky. I guess it's due to the fact that this job installed NumPy 1.x where copy signature is bool | _CopyMode. Locally I see that installing Numpy 2.0dev yields even more typing errors in other places (like error: Name "np.bool_" is not defined as bool_ typing stub got renamed to bool).

So for this PR, maybe # type: ignore for these calls would work?

error: Argument "copy" to "array" has incompatible type "bool | None"; expected "bool | _CopyMode"  [arg-type]

[jorisvandenbossche]

Suggested change

	copy_false = None if np_version_gt2 else False
	result = np.array(result, copy=copy_false)
	result = np.asarray(result)

Would this be a simpler but equivalent alternative?

[rgommers]

Looks like it (from a quick glance at the diff).

I just added this to the 2.0 migration guide:

1. Code using np.array(..., copy=False) can in most cases be changed to
   np.asarray(...). Older code tended to use np.array like this because
   it had less overhead than the default np.asarray copy-if-needed
   behavior. This is no longer true, and np.asarray is the preferred function.

[lumberbot-app]

Owee, I'm MrMeeseeks, Look at me.

There seem to be a conflict, please backport manually. Here are approximate instructions:

1. Checkout backport branch and update it.

git checkout 2.2.x
git pull

2. Cherry pick the first parent branch of the this PR on top of the older branch:

git cherry-pick -x -m1 b89f1d0d05f4c9f360985abc6bda421d73bae85f

3. You will likely have some merge/cherry-pick conflict here, fix them and commit:

git commit -am "Backport PR #57172: MAINT: Adjust the codebase to the new `np.array`'s `copy` keyword meaning"

4. Push to a named branch:

git push YOURFORK 2.2.x:auto-backport-of-pr-57172-on-2.2.x

5. Create a PR against branch 2.2.x, I would have named this PR:

"Backport PR #57172 on branch 2.2.x (MAINT: Adjust the codebase to the new np.array's copy keyword meaning)"

And apply the correct labels and milestones.

Congratulations — you did some good work! Hopefully your backport PR will be tested by the continuous integration and merged soon!

Remember to remove the Still Needs Manual Backport label once the PR gets merged.

If these instructions are inaccurate, feel free to suggest an improvement.

[mroeschke]

Thanks @mtsokol. Will create a follow up issue about utilizing the new copy keyword in __array__