Overview

1 Release published by1 person

• v2.2.6 v2.2.6 (May 17, 2025)

  publishedMay 17, 2025

30 Pull requests merged by17 people

• DOC: Expand/clean up extension module import error

  #29036 mergedMay 23, 2025

• CI: update cibuildwheel to 3.0.0b1 and enable cp314 and cp314t nightlies

  #29007 mergedMay 23, 2025

• BUG: Fix cache use regression

  #29046 mergedMay 23, 2025

• BUG: Fix f2py derived types in modules

  #29044 mergedMay 23, 2025

• BUG: Avoid compile errors in f2py modules

  #29043 mergedMay 23, 2025

• MNT: constant string arrays instead of pointers in C

  #29042 mergedMay 23, 2025

• BUG: Fix cache use regression

  #29041 mergedMay 23, 2025

• ENH: add __array_function__ protocol in polynomial

  #28996 mergedMay 23, 2025

• BUG: Fix f2py derived types in modules

  #29040 mergedMay 23, 2025

• BUG: Avoid compile errors in f2py modules

  #29034 mergedMay 23, 2025

• MNT: constant string arrays instead of pointers in C

  #28985 mergedMay 23, 2025

• BUG: Fix workflow error

  #29033 mergedMay 21, 2025

• BUG: Fix workflow bug

  #29032 mergedMay 21, 2025

• BEG, MAINT: Begin NumPy 2.4.0 development.

  #29019 mergedMay 21, 2025

• MAINT: Bump larsoner/circleci-artifacts-redirector-action from 1.0.0 to 1.1.0

  #29018 mergedMay 21, 2025

• DOC: Fixes absent line numbers on link to classes decorated with set_module

  #28918 mergedMay 20, 2025

• MNT: fix CI issues on main

  #29008 mergedMay 20, 2025

• TYP: TypeMaskedArray.__{iadd,isub,imul,itruediv,ifloordiv,ipow}__

  #28986 mergedMay 19, 2025

• ENH: Disable the alloc cache under address and memory sanitizers

  #29006 mergedMay 19, 2025

• TYP: allow inplace division ofNDArray[timedelta64] by floats

  #28994 mergedMay 19, 2025

• TYP: Gradual shape type defaults

  #28982 mergedMay 19, 2025

• BUG: allclose does not warn for invalid value encountered in multiply or alternativly checks for validity of tolerances

  #28221 mergedMay 19, 2025

• MAINT: Update main after 2.2.6 release.

  #28998 mergedMay 19, 2025

• MAINT: Update download-wheels for multiple pages

  #29002 mergedMay 19, 2025

• MAINT: update NPY_FEATURE_VERSION after dropping python 3.10

  #28993 mergedMay 18, 2025

• MAINT: Prepare 2.2.x for further development

  #28997 mergedMay 17, 2025

• TYP: removefrom __future__ import annotations

  #28995 mergedMay 17, 2025

• Revert "DEP: Deprecate.T property for non-2dim arrays and scalars"

  #28990 mergedMay 16, 2025

• DEP: Deprecate.T property for non-2dim arrays and scalars

  #28678 mergedMay 16, 2025

• TYP: add missingndarray.__{add,mul}__character type overloads

  #28988 mergedMay 16, 2025

14 Pull requests opened by5 people

• ENH: Convert minmax from C universal intrinsics to C++ using Highway

  #29001 openedMay 18, 2025

• TYP: Type ``MaskedArray.__{add,radd,sub,rsub}__``

  #29012 openedMay 20, 2025

• MNT: xfail flaky test on pypy

  #29014 openedMay 20, 2025

• GH-29021: Convert multiarray to multi-phase init (PEP 489)

  #29022 openedMay 21, 2025

• GH-29021: Convert umath to multi-phase init (PEP 489)

  #29023 openedMay 21, 2025

• GH-29021: Convert the ufunc tutorial to multi-phase init (PEP 489)

  #29024 openedMay 21, 2025

• GH-29021: Convert _simd.c to multi-phase init (PEP 489)

  #29025 openedMay 21, 2025

• GH-29021: Convert dummymodule to multi-phase init (PEP 489)

  #29026 openedMay 21, 2025

• GH-29021: Convert limited API examples to multi-phase init (PEP 489)

  #29027 openedMay 21, 2025

• GH-29021: Convert pocketfft_umath to multi-phase init (PEP 489)

  #29028 openedMay 21, 2025

• GH-29021: Convert lapack_lite to multi-phase init (PEP 489)

  #29029 openedMay 21, 2025

• GH-29021: Convert umath_linalg to multi-phase init (PEP 489)

  #29030 openedMay 21, 2025

• GH-29021: Convert extbuild to multi-phase init (PEP 489)

  #29031 openedMay 21, 2025

• BLD: bump OpenBLAS version, use OpenBLAS for win-arm64

  #29039 openedMay 23, 2025

10 Issues closed by7 people

• BUG: f2py failing to parse fix-form Fortran

  #28749 closedMay 23, 2025

• ENH: add __array_function__ protocol to polyval2d and polygrid2d

  #28866 closedMay 23, 2025

• BUG: f2py fails to compile modules with derived types after changes in 2.2.0

  #29037 closedMay 23, 2025

• BUG: cannot use `np.save` and `allow_pickle=True` with data larger than 4 GB

  #26224 closedMay 22, 2025

• The windows actions test does not seem to be run.

  #29020 closedMay 21, 2025

• ENH: Multithreaded sort / x86-simd-sort

  #29009 closedMay 20, 2025

• int64 is not supported by the standard library's json encoder

  #27651 closedMay 20, 2025

• BUG: numpy gradient yields incorrect results

  #28999 closedMay 18, 2025

• Update NPY_FEATURE_VERSION after dropping Python 3.10

  #28992 closedMay 18, 2025

• BUG: C23 Compiler Flags Ignored or Obscured During NumPy Build via Meson Backend

  #28953 closedMay 17, 2025

10 Issues opened by10 people

• BUG: `np.nan_to_num(..., copy=False)` raises `ValueError` in NumPy 2.0 when array creation requires copy

  #29045 openedMay 23, 2025

• BUG: NumPy 2.3.0 regression - corrupted array values

  #29038 openedMay 23, 2025

• BLD: use OpenBLAS in the windows-arm64 build

  #29035 openedMay 22, 2025

• ENH: Use multi-phase initialisation (PEP 489) for extension modules

  #29021 openedMay 21, 2025

• No module named 'numpy.core._multiarray_umath' ~ Bazel Edition

  #29017 openedMay 21, 2025

• BUG: in-place fixed-width string multiply doesn't do overflow checking

  #29011 openedMay 20, 2025

• Packaging openblas libraries with numpy wheel

  #29010 openedMay 20, 2025

• BUG: np.percentile fails with internal overflow when using float16 input on large arrays

  #29003 openedMay 19, 2025

• Make `eig/eigvals` always return complex eigenvalues

  #29000 openedMay 18, 2025

• GCC 15.1.1 compiles NumPy, but the tests segfault.

  #28991 openedMay 16, 2025

32 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• ENH: Modulate dispatched x86 CPU features

  #28896 commented onMay 19, 2025 • 10 new comments

• ENH: Convert unary_complex from C universal intrinsics to C++ using Highway

  #28815 commented onMay 21, 2025 • 3 new comments

• ENH: ``__trunc__`` for floating and integer types

  #28949 commented onMay 19, 2025 • 2 new comments

• ENH, API: New sorting mechanism for DType API

  #28516 commented onMay 22, 2025 • 2 new comments

• BUG: quantile should error when weights are all zeros

  #28595 commented onMay 21, 2025 • 1 new comment

• DOC: Fix ambiguity in polyfit description

  #28147 commented onMay 23, 2025 • 1 new comment

• ENH: np.unique: support hash based unique for string dtype

  #28767 commented onMay 22, 2025 • 1 new comment

• DOC: Add examples to broadcasting documentation for 2D and 3D array cases

  #28891 commented onMay 23, 2025 • 0 new comments

• ENH: Convert unary_fp_le from C universal intrinsics to C++ using Highway

  #28919 commented onMay 18, 2025 • 0 new comments

• BUG: Converting large integer to np.longdouble can raise ValueError

  #28722 commented onMay 22, 2025 • 0 new comments

• ENH: Convert unary_fp from C universal intrinsics to C++ using Highway

  #28718 commented onMay 18, 2025 • 0 new comments

• BUG: Respect print options in masked array formatting

  #28698 commented onMay 16, 2025 • 0 new comments

• ENH, SIMD: Initial implementation of Highway wrapper

  #28622 commented onMay 20, 2025 • 0 new comments

• ENH: Convert unary from C universal intrinsics to C++ using Highway

  #28608 commented onMay 20, 2025 • 0 new comments

• DEP: Deprecate setting the strides attribute of a numpy array

  #28925 commented onMay 18, 2025 • 0 new comments

• DOC: Clarify (potentially misleading) nbytes docstring

  #28943 commented onMay 20, 2025 • 0 new comments

• DOC: improves np.fromfile file description (#28840)

  #28979 commented onMay 20, 2025 • 0 new comments

• ENH: Use array indexing preparation routines for flatiter objects

  #28590 commented onMay 19, 2025 • 0 new comments

• BUG: TypeError: add_docstring() argument 2 must be str, not None

  #13248 commented onMay 23, 2025 • 0 new comments

• ENH: assert_allclose show indices and values of maximum differences on failure

  #28827 commented onMay 23, 2025 • 0 new comments

• BUG: wrong errors e.g. "divide by zero encountered in matmul" on MacOS M4

  #28687 commented onMay 23, 2025 • 0 new comments

• DOC: Add information (maybe just release note) for typing regressions

  #28077 commented onMay 22, 2025 • 0 new comments

• Formal names for 2.3.0 contributors

  #28951 commented onMay 22, 2025 • 0 new comments

• ENH: Should row_stack really be deprecated?

  #26032 commented onMay 21, 2025 • 0 new comments

• TSK: deprecate setting `strides`, `dtype`, and `shape` on arrays.

  #28800 commented onMay 20, 2025 • 0 new comments

• BUG: scalars missing several methods for array api compat

  #27305 commented onMay 20, 2025 • 0 new comments

• BUG: scalars don't have dlpack support

  #27137 commented onMay 20, 2025 • 0 new comments

• `where` with out-of-bounds Python integers in Numpy 2

  #27006 commented onMay 20, 2025 • 0 new comments

• ENH: expose numpy.dtype.flags bitmasks in numpy.dtypes

  #24963 commented onMay 20, 2025 • 0 new comments

• BUG: Polynomial package slower than polynomial module for python <= 3.11. Both are worse for python > 3.11.

  #28948 commented onMay 20, 2025 • 0 new comments

• Avoid copy in matrix-vector dots with negative strides?

  #28909 commented onMay 18, 2025 • 0 new comments

• moving x86-64 feature baseline to SSE4.2?

  #27851 commented onMay 17, 2025 • 0 new comments