Drop support for numpy 1.x by mhvk · Pull Request #18986 · astropy/astropy

mhvk · 2025-11-24T22:33:11Z

This drops support for numpy < 2.0. It is in a few commits:

Oversight from a long time ago.
Drop numpy 1.25 (very few changes).
Drop numpy 1.26 (lots of changes).
Also remove COPY_IF_NEEDED, since it is now always just None (lots of changes).

EDIT: further commits
5. Bump other dependencies to the minimum needed for numpy >=2.0 support (inferred from numpy/numpy#26191)
6. Silence a weird RuntimeError (happens on main too with same python/numpy combo; see #18989.

In total ~500 lines changed and ~600 lines deleted, i.e., lots of compatibility work-arounds simplified.

By checking this box, the PR author has requested that maintainers do NOT use the "Squash and Merge" button. Maintainers should respect this when possible; however, the final decision is at the discretion of the maintainer that merges the PR.

github-actions · 2025-11-24T22:33:24Z

Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.

Do the proposed changes actually accomplish desired goals?
Do the proposed changes follow the Astropy coding guidelines?
Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see instructions for rebase and squash.
Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the "Extra CI" label. Codestyle issues can be fixed by the bot.
Is a change log needed? If yes, did the change log check pass? If no, add the "no-changelog-entry-needed" label. If this is a manual backport, use the "skip-changelog-checks" label unless special changelog handling is necessary.
Is this a big PR that makes a "What's new?" entry worthwhile and if so, is (1) a "what's new" entry included in this PR and (2) the "whatsnew-needed" label applied?
At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate "backport-X.Y.x" label(s) before merge.

pllim · 2025-11-24T22:46:51Z

Wow, @mhvk , you are brave...

If you push https://github.com/astropy/astropy/releases/tag/v8.0.0.dev tag to your fork, I can try to run your PR branch here in https://github.com/astropy/astropy-integration-testing/ and see how many things break.

neutrinoceros

Here a complete review with a handful of nits, but overall nothing major to report.

neutrinoceros · 2026-01-01T12:01:38Z

-                            warnings.simplefilter("ignore", category=DeprecationWarning)
-                            arr.data = self[idx].data.data
-                    elif sys.getrefcount(arr) > 2:
+                    if sys.getrefcount(arr) > 2:


minor: we could now save one indentation level in what follows by using a single guard clause (if mmap is None or sys.getrefcount(arr) <= 2)

neutrinoceros · 2026-01-01T12:06:25Z

+DTYPES_TEST_INIT = ["?", "b", "i2", "f4", "c8", "S", "U", "O"] + (
+    [  # new in numpy 2
+        np.dtypes.StrDType,  # same as "U"
+        np.dtypes.BytesDType,  # same as "S"
+        np.dtypes.StringDType,
+    ]
+)


the extra parens are somewhat distracting and can be removed

neutrinoceros · 2026-01-01T12:13:34Z


 if NUMPY_LT_2_4:
-    # np.in1d deprecated in not NUMPY_LT_2_0, removed in not NUMPY_LT_24
+    # np.in1d removed in not NUMPY_LT_24


whoops

Suggested change

# np.in1d removed in not NUMPY_LT_24

# np.in1d removed in not NUMPY_LT_2_4

Rephrased as text, since it is now a simple if statement so no need for being greppable.

neutrinoceros · 2026-01-01T12:19:23Z

            obj = self.view(np.ndarray)
        else:
-            obj = np.array(obj, copy=COPY_IF_NEEDED, subok=True)
+            obj = np.array(obj, copy=None, subok=True)


for my own culture, is this any different than the following ?

Suggested change

obj = np.array(obj, copy=None, subok=True)

obj = np.asanyarray(obj, copy=None)

It it the same, but the np.array call used to be faster, since np.asanyarray would call np.array... But in trying to proof that, I found it has actually changed: np.asanyarray now is faster: with a plain array as input, which is most likely here, it is 75 vs 85 ns on my laptop, and down to 39 ns after removing copy=None (which is the default for np.asanyarray). So, I'm changing it!

neutrinoceros · 2026-01-01T12:23:36Z

-                    ) = np.lib._function_base_impl._parse_gufunc_signature(
-                        ufunc.signature.replace(" ", "")
-                    )
+                (in_sig, out_sig) = np.lib._function_base_impl._parse_gufunc_signature(


parens serve no purpose now

Suggested change

(in_sig, out_sig) = np.lib._function_base_impl._parse_gufunc_signature(

in_sig, out_sig = np.lib._function_base_impl._parse_gufunc_signature(

neutrinoceros · 2026-01-01T12:28:48Z

    # a direct dependency in order to set a lower bound for oldest-deps testing
    "jupyter-core>=4.11.2",
-    "pandas>=2.0.2",
+    "pandas>=2.2.2",  # minimum that supports numpy 2.0


same here

Suggested change

"pandas>=2.2.2", # minimum that supports numpy 2.0

"pandas>=2.2.2",

OK, also done.

neutrinoceros · 2026-01-01T12:29:33Z

-    "pyarrow>=14.0.2",
-    "beautifulsoup4>=4.9.3", # imposed by pandas==2.0
+    "h5py>=3.11.0",  # older versions are not compatible with numpy 2.0
+    "pyarrow>=16.0",  # older versions are not compatible with numpy 2.0


Suggested change

"pyarrow>=16.0", # older versions are not compatible with numpy 2.0

"h5py>=3.11.0",

"pyarrow>=16.0",

neutrinoceros · 2026-01-01T12:29:45Z

    "mpmath>=1.2.1",
    "asdf-astropy>=0.7.0",
-    "bottleneck>=1.3.3",
+    "bottleneck>=1.4.0",  # minimum for numpy 2.0 compatibility


Suggested change

"bottleneck>=1.4.0", # minimum for numpy 2.0 compatibility

"bottleneck>=1.4.0",

OK, those are indeed all in one commit, so easy enough to find if needed.

neutrinoceros · 2026-01-01T12:30:10Z

+    "pyarrow>=16.0",  # minimum for numpy 2.0 support
+    "pandas>=2.2.2",  # minimum for numpy 2.0 support


Suggested change

"pyarrow>=16.0", # minimum for numpy 2.0 support

"pandas>=2.2.2", # minimum for numpy 2.0 support

"pyarrow>=16.0",

"pandas>=2.2.2",

neutrinoceros · 2026-01-01T12:32:22Z


 deps =
-    numpy125: numpy==1.25.*
+    numpy20: numpy==2.0.*


I would actually suggest dropping this tox env now; I don't think it's used anywhere and it's not providing much value beyond oldestdeps

I agree with this comment.

OK, done -- there defintely are more than enough environments as is, and this one has outlived its purpose.

neutrinoceros · 2026-01-01T12:38:20Z

Before I reviewed I also pushed a commit to change mpl380 to mpl384, but seeing that it makes for a more painful merge, I moved the commit to #19115

pllim · 2026-01-01T22:16:35Z

Would it help if I do the mpl384 bump as a standalone PR when I return?

neutrinoceros · 2026-01-01T23:44:03Z

That's already the content of #19115, and yes, it can wait for you :)

pllim · 2026-01-02T14:43:18Z

@neutrinoceros , is it possible to take this PR out commits out of #19115 then? mpl upgrade on CircleCI should be completely separable from numpy upgrade.

neutrinoceros · 2026-01-02T14:53:54Z

Indeed. It'll create a couple trivial merge conflict but as long as it helps review, that's fine with me. Done.

This warning is not specific to this PR, but happens on generally on numpy 2.0 or 2.1 and python 3.10, 3.11 (maybe later too). It was not seen before because CI did not test the relevant combinations.

mhvk · 2026-01-02T18:42:49Z

@neutrinoceros - I implemented all your suggestions (all clear improvements), thanks!

neutrinoceros

Thank you ! I trust you did go through all of them and I don't think I need a second full scan. Approving now !

mhvk · 2026-01-02T21:12:59Z

Thanks, merging this...

mhvk added this to the v8.0.0 milestone Nov 24, 2025

mhvk requested a review from neutrinoceros as a code owner November 24, 2025 22:33

mhvk added the dependencies Pull requests that update a dependency file label Nov 24, 2025

mhvk requested review from a team, WilliamJamieson, larrybradley, matteobachetti, saimn and taldcroft as code owners November 24, 2025 22:33

mhvk requested a review from nstarman November 24, 2025 22:33

github-actions Bot added testing table io.fits time nddata units coordinates modeling cosmology utils io.misc installation stats uncertainty utils.masked labels Nov 24, 2025

github-project-automation Bot added this to Cosmology, the Expansion Nov 24, 2025

github-project-automation Bot moved this to Triage in Cosmology, the Expansion Nov 24, 2025

neutrinoceros requested changes Jan 1, 2026

View reviewed changes

neutrinoceros mentioned this pull request Jan 1, 2026

TST: replace mpl380 tox factor with oldestdeps and solidify reproducibility of image tests #19115

Closed

1 task

neutrinoceros force-pushed the drop-numpy-1.x branch from f50b5bc to e9cf40d Compare January 1, 2026 12:38

This comment was marked as duplicate.

Sign in to view

mhvk added 10 commits January 2, 2026 13:40

MAINT: instead of xfail for numpy >=1.10, expect error.

6024f27

MAINT: bump minimum numpy version to 1.26

2004498

MAINT: Bump minimum version of numpy to 2.0

68cfc72

MAINT: replace COPY_IF_NEEDED with None

083599b

MAINT: deprecate, not remove COPY_IF_NEEDED

e6180fe

MAINT: bump other packages for compatibility with numpy 2.0

ab92ec8

MAINT: ignore warning from Angle.to_string() for now

56b2fe8

This warning is not specific to this PR, but happens on generally on numpy 2.0 or 2.1 and python 3.10, 3.11 (maybe later too). It was not seen before because CI did not test the relevant combinations.

MAINT: update minimum asdf, scipy, pyerfa for oldest-deps

ae6fde9

MAINT: update minimum required matplotlib to 3.8.4

275ac20

MAINT: Cleanups suggested by Clément Robert

4621aea

mhvk force-pushed the drop-numpy-1.x branch from e9cf40d to 4621aea Compare January 2, 2026 18:42

neutrinoceros approved these changes Jan 2, 2026

View reviewed changes

github-project-automation Bot moved this from Review in progress to Reviewer approved in Cosmology, the Expansion Jan 2, 2026

mhvk merged commit 7f29b3e into astropy:main Jan 2, 2026
39 of 40 checks passed

github-project-automation Bot moved this from Reviewer approved to Done in Cosmology, the Expansion Jan 2, 2026

mhvk deleted the drop-numpy-1.x branch January 2, 2026 21:13

rosteen mentioned this pull request Jan 5, 2026

Avoid deprecation warning with dev astropy/astropy>=8 radio-astro-tools/spectral-cube#991

Merged

neutrinoceros mentioned this pull request Apr 29, 2026

Backport PR #19610 to v7.2.x (TST: Add back ignoring socket warnings of yore) #19615

Merged

1 task

neutrinoceros mentioned this pull request Sep 25, 2026

BLD: update NPY_* definitions to match runtime requirements #20481

Merged

1 of 2 tasks

	# np.in1d removed in not NUMPY_LT_24
	# np.in1d removed in not NUMPY_LT_2_4

	obj = np.array(obj, copy=None, subok=True)
	obj = np.asanyarray(obj, copy=None)

	(in_sig, out_sig) = np.lib._function_base_impl._parse_gufunc_signature(
	in_sig, out_sig = np.lib._function_base_impl._parse_gufunc_signature(

	"pandas>=2.2.2", # minimum that supports numpy 2.0
	"pandas>=2.2.2",

	"pyarrow>=16.0", # older versions are not compatible with numpy 2.0
	"h5py>=3.11.0",
	"pyarrow>=16.0",

	"bottleneck>=1.4.0", # minimum for numpy 2.0 compatibility
	"bottleneck>=1.4.0",

		"pyarrow>=16.0", # minimum for numpy 2.0 support
		"pandas>=2.2.2", # minimum for numpy 2.0 support

Conversation

mhvk commented Nov 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Nov 24, 2025

Uh oh!

pllim commented Nov 24, 2025

Uh oh!

neutrinoceros left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

neutrinoceros commented Jan 1, 2026

Uh oh!

pllim commented Jan 1, 2026

Uh oh!

neutrinoceros commented Jan 1, 2026

Uh oh!

This comment was marked as duplicate.

pllim commented Jan 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

neutrinoceros commented Jan 2, 2026

Uh oh!

mhvk commented Jan 2, 2026

Uh oh!

neutrinoceros left a comment

Choose a reason for hiding this comment

Uh oh!

mhvk commented Jan 2, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

9 participants

mhvk commented Nov 24, 2025 •

edited

Loading

pllim commented Jan 2, 2026 •

edited

Loading