NumPy 2.3.4 Release Notes

The NumPy 2.3.4 release is a patch release split between a number of maintenance
updates and bug fixes. This release supports Python versions 3.11-3.14. This
release is based on NumPy 3.14.0 final.

Changes

The npymath and npyrandom libraries now have a .lib rather than a
.a file extension on win-arm64, for compatibility for building with MSVC and
setuptools. Please note that using these static libraries is discouraged
and for existing projects using it, it's best to use it with a matching
compiler toolchain, which is clang-cl on Windows on Arm.

(gh-29750)

Contributors

A total of 17 people contributed to this release. People with a "+" by their
names contributed a patch for the first time.

• !DWesl
• Charles Harris
• Christian Barbia +
• Evgeni Burovski
• Joren Hammudoglu
• Maaz +
• Mateusz Sokół
• Matti Picus
• Nathan Goldbaum
• Ralf Gommers
• Riku Sakamoto +
• Sandeep Gupta +
• Sayed Awad
• Sebastian Berg
• Sergey Fedorov +
• Warren Weckesser
• dependabot[bot]

Pull requests merged

A total of 30 pull requests were merged for this release.

• #29725: MAINT: Prepare 2.3.x for further development
• #29781: MAINT: Pin some upstream dependences
• #29782: BLD: enable x86-simd-sort to build on KNL with -mavx512f
• #29783: BUG: Include python-including headers first (#29281)
• #29784: TYP: fix np.number and np.*integer method declaration
• #29785: TYP: mypy 1.18.1
• #29788: TYP: replace scalar type __init__ with __new__
• #29790: BUG: Fix dtype refcount in __array__ (#29715)
• #29791: TYP: fix method declarations in floating, timedelta64, and datetime64Backport
• #29792: MAINT: delete unused variables in unary logical dispatch
• #29797: BUG: Fix pocketfft umath strides for AIX compatibility (#29768)
• #29798: BUG: np.setbufsize should raise ValueError for negative input
• #29799: BUG: Fix assert in nditer buffer setup
• #29800: BUG: Stable ScalarType ordering
• #29838: TST: Pin pyparsing to avoid matplotlib errors.
• #29839: BUG: linalg: emit a MemoryError on a malloc failure (#29811)
• #29840: BLD: change file extension for libnpymath on win-arm64 from .a...
• #29864: CI: Fix loongarch64 CI (#29856)
• #29865: TYP: Various typing fixes
• #29910: BUG: Fix float16-sort failures on 32-bit x86 MSVC (#29908)
• #29911: TYP: add missing __slots__ (#29901)
• #29913: TYP: wrong argument defaults in testing._private (#29902)
• #29920: BUG: avoid segmentation fault in string_expandtabs_length_promoter
• #29921: BUG: Fix INT_MIN % -1 to return 0 for all signed integer types...
• #29922: TYP: minor fixes related to errstate (#29914)
• #29923: TST: use requirements/test_requirements across CI (#29919)
• #29926: BUG: fix negative samples generated by Wald distribution (#29609)
• #29940: MAINT: Bump pypa/cibuildwheel from 3.1.4 to 3.2.1
• #29949: STY: rename @classmethod arg to cls
• #29950: MAINT: Simplify string arena growth strategy (#29885)