From 42edc1a9632b3e2579646d26a108c7b2c8d19b03 Mon Sep 17 00:00:00 2001 From: Charles Harris Date: Wed, 7 May 2025 21:16:50 -0600 Subject: [PATCH] REL: Prepare for the NumPy 2.3.0rc1 release [wheel build] - Create 2.3.0-changelog.rst - Update 2.3.0-notes.rst - Update pyproject.toml - Update .mailmap - Delete release note fragments --- .mailmap | 34 +- doc/changelog/2.3.0-changelog.rst | 689 ++++++++++++++++++ doc/release/upcoming_changes/26018.change.rst | 7 - .../upcoming_changes/26018.performance.rst | 7 - .../upcoming_changes/26745.highlight.rst | 10 - .../upcoming_changes/27288.improvement.rst | 3 - .../upcoming_changes/27789.new_function.rst | 5 - doc/release/upcoming_changes/27883.c_api.rst | 4 - doc/release/upcoming_changes/27883.change.rst | 17 - doc/release/upcoming_changes/27998.c_api.rst | 10 - doc/release/upcoming_changes/28080.c_api.rst | 1 - .../upcoming_changes/28080.improvement.rst | 2 - doc/release/upcoming_changes/28102.change.rst | 6 - .../upcoming_changes/28105.improvement.rst | 2 - .../upcoming_changes/28129.deprecation.rst | 4 - .../upcoming_changes/28205.improvement.rst | 6 - .../upcoming_changes/28214.new_feature.rst | 23 - .../upcoming_changes/28250.improvement.rst | 3 - .../upcoming_changes/28254.expired.rst | 29 - doc/release/upcoming_changes/28343.change.rst | 1 - doc/release/upcoming_changes/28426.change.rst | 6 - doc/release/upcoming_changes/28436.change.rst | 10 - .../upcoming_changes/28442.improvement.rst | 1 - doc/release/upcoming_changes/28569.change.rst | 2 - .../upcoming_changes/28576.new_feature.rst | 15 - doc/release/upcoming_changes/28615.change.rst | 5 - .../upcoming_changes/28619.highlight.rst | 6 - .../upcoming_changes/28619.performance.rst | 7 - .../upcoming_changes/28669.new_feature.rst | 3 - doc/release/upcoming_changes/28703.change.rst | 3 - doc/release/upcoming_changes/28713.change.rst | 1 - doc/release/upcoming_changes/28741.change.rst | 1 - .../upcoming_changes/28769.performance.rst | 8 - .../upcoming_changes/28856.improvement.rst | 5 - .../upcoming_changes/28884.deprecation.rst | 28 - .../upcoming_changes/28940.new_feature.rst | 6 - .../upcoming_changes/28961.expired.rst | 1 - doc/source/release/2.3.0-notes.rst | 513 ++++++++++++- pyproject.toml | 2 +- 39 files changed, 1231 insertions(+), 255 deletions(-) create mode 100644 doc/changelog/2.3.0-changelog.rst delete mode 100644 doc/release/upcoming_changes/26018.change.rst delete mode 100644 doc/release/upcoming_changes/26018.performance.rst delete mode 100644 doc/release/upcoming_changes/26745.highlight.rst delete mode 100644 doc/release/upcoming_changes/27288.improvement.rst delete mode 100644 doc/release/upcoming_changes/27789.new_function.rst delete mode 100644 doc/release/upcoming_changes/27883.c_api.rst delete mode 100644 doc/release/upcoming_changes/27883.change.rst delete mode 100644 doc/release/upcoming_changes/27998.c_api.rst delete mode 100644 doc/release/upcoming_changes/28080.c_api.rst delete mode 100644 doc/release/upcoming_changes/28080.improvement.rst delete mode 100644 doc/release/upcoming_changes/28102.change.rst delete mode 100644 doc/release/upcoming_changes/28105.improvement.rst delete mode 100644 doc/release/upcoming_changes/28129.deprecation.rst delete mode 100644 doc/release/upcoming_changes/28205.improvement.rst delete mode 100644 doc/release/upcoming_changes/28214.new_feature.rst delete mode 100644 doc/release/upcoming_changes/28250.improvement.rst delete mode 100644 doc/release/upcoming_changes/28254.expired.rst delete mode 100644 doc/release/upcoming_changes/28343.change.rst delete mode 100644 doc/release/upcoming_changes/28426.change.rst delete mode 100644 doc/release/upcoming_changes/28436.change.rst delete mode 100644 doc/release/upcoming_changes/28442.improvement.rst delete mode 100644 doc/release/upcoming_changes/28569.change.rst delete mode 100644 doc/release/upcoming_changes/28576.new_feature.rst delete mode 100644 doc/release/upcoming_changes/28615.change.rst delete mode 100644 doc/release/upcoming_changes/28619.highlight.rst delete mode 100644 doc/release/upcoming_changes/28619.performance.rst delete mode 100644 doc/release/upcoming_changes/28669.new_feature.rst delete mode 100644 doc/release/upcoming_changes/28703.change.rst delete mode 100644 doc/release/upcoming_changes/28713.change.rst delete mode 100644 doc/release/upcoming_changes/28741.change.rst delete mode 100644 doc/release/upcoming_changes/28769.performance.rst delete mode 100644 doc/release/upcoming_changes/28856.improvement.rst delete mode 100644 doc/release/upcoming_changes/28884.deprecation.rst delete mode 100644 doc/release/upcoming_changes/28940.new_feature.rst delete mode 100644 doc/release/upcoming_changes/28961.expired.rst diff --git a/.mailmap b/.mailmap index f33dfddb6492..3fde6c379026 100644 --- a/.mailmap +++ b/.mailmap @@ -65,6 +65,8 @@ !yetanothercheer Aaron Baecker Adrin Jalali +Abhishek Kumar +Abhishek Kumar <142383124+abhishek-iitmadras@users.noreply.github.com> Abraham Medina Arun Kota Arun Kota Arun Kota @@ -140,6 +142,8 @@ Anton Prosekin Anže Starič Arfy Slowy Arnaud Ma +Arnaud Tremblay +Arnaud Tremblay <59627629+Msa360@users.noreply.github.com> Aron Ahmadia Arun Kota Arun Kota @@ -190,6 +194,8 @@ Carl Kleffner Carl Leake Carlos Henrique Hermanny Moreira da Silva Carlos Henrique Hermanny Moreira da Silva <121122527+carlosilva10260@users.noreply.github.com> +Carlos Martin +Carlos Martin Cédric Hannotier Charles Stern <62192187+cisaacstern@users.noreply.github.com> Chiara Marmo @@ -200,6 +206,8 @@ Chris Burns Chris Fu (傅立业) <17433201@qq.com> Chris Holland <41524756+ChrisAHolland@users.noreply.github.com> Chris Kerr +Chris Navarro +Chris Navarro <24905907+lvllvl@users.noreply.github.com> Chris Vavaliaris Christian Clauss Christopher Dahlin @@ -270,6 +278,7 @@ Eric Fode Eric Fode Eric Quintero Eric Xie <161030123+EngineerEricXie@users.noreply.github.com> Ernest N. Mamikonyan +Ernst Peng Eskild Eriksen Eskild Eriksen <42120229+iameskild@users.noreply.github.com> Eskild Eriksen @@ -300,8 +309,11 @@ Gregory R. Lee Gregory R. Lee Guo Ci guoci Guo Shuai +Gyeongjae Choi Habiba Hye Habiba Hye <145866308+HabibiHye@users.noreply.github.com> +Halle Loveday +Halle Loveday Hameer Abbasi Hannah Aizenman Han Genuit @@ -311,6 +323,10 @@ Helder Oliveira Hemil Desai Himanshu Hiroyuki V. Yamazaki +Daniel Hrisca +Daniel Hrisca +François de Coatpont +François de Coatpont <93073405+Chevali2004@users.noreply.github.com> Hugo van Kemenade Iantra Solari I-Shen Leong @@ -363,6 +379,7 @@ Jessé Pires Jessi J Zhao <35235453+jessijzhao@users.noreply.github.com> Jhong-Ken Chen (陳仲肯) Jhong-Ken Chen (陳仲肯) <37182101+kennychenfs@users.noreply.github.com> +Jiuding Tan (谭九鼎) <109224573@qq.com> Johann Faouzi Johann Rohwer Johann Rohwer jmrohwer @@ -447,10 +464,13 @@ Luke Zoltan Kelley Madhulika Jain Chambers <53166646+madhulikajc@users.noreply.github.com> Magdalena Proszewska Magdalena Proszewska <38814059+mproszewska@users.noreply.github.com> +Makima C. Yang Malik Idrees Hasan Khan <77000356+MalikIdreesHasanKhan@users.noreply.github.com>C Manoj Kumar Marcel Loose Marcin Podhajski <36967358+m-podhajski@users.noreply.github.com> +Marco Edward Gorelli +Marco Edward Gorelli <33491632+MarcoGorelli@users.noreply.github.com> Margret Pax Margret Pax <13646646+paxcodes@users.noreply.github.com> Mark DePristo @@ -508,6 +528,7 @@ Michael Schnaitter Michael Seifert Michel Fruchart +Mike O'Brien Mike Toews Miki Watanabe (渡邉 美希) Miles Cranmer @@ -516,9 +537,12 @@ Milica Dančuk love-bees <33499899+love-bees@users.noreply.g Mircea Akos Bruma Mircea Akos Bruma Mitchell Faas <35742861+Mitchell-Faas@users.noreply.github.com> +Mohammed Abdul Rahman +Mohammed Abdul Rahman <130785777+that-ar-guy@users.noreply.github.com> Mohaned Qunaibit Muhammad Kasim Muhammed Muhsin +Mugundan Selvanayagam Mukulika Pahari Mukulika Pahari <60316606+Mukulikaa@users.noreply.github.com> Munira Alduraibi @@ -571,6 +595,7 @@ Peter J Cock Peter Kämpf Peyton Murray Phil Elson +Filipe Laíns Pierre GM Pierre GM pierregm Piotr Gaiński @@ -591,6 +616,8 @@ Rehas Sachdeva Richard Howe <45905457+rmhowe425@users.noreply.github.com> Ritta Narita Riya Sharma +Rob Timpe +Rob Timpe Robert Kern Robert LU Robert T. McGibbon @@ -660,6 +687,7 @@ Steve Stagg Steven J Kern Stuart Archibald Stuart Archibald +SUMIT SRIMANI <2301109104@ptuniv.edu.in SuryaChand P Sylvain Ferriol Takanori Hirano @@ -696,6 +724,8 @@ Vinith Kishore <85550536+vinith2@users.noreply.github Vrinda Narayan Vrinda Narayan Vrinda Narayan <48102157+vrindaaa@users.noreply.github.com> +Wang Yang (杨旺) +Wang Yang (杨旺) <1113177880@qq.com> Wansoo Kim Warrick Ball Warrick Ball @@ -711,11 +741,11 @@ Xiangyi Wang Yamada Fuyuka Yang Hau Yang Hau -Yang Wang Yash Pethe Yash Pethe <83630710+patient74@users.noreply.github.com> Yashasvi Misra Yashasvi Misra <54177363+yashasvimisra2798@users.noreply.github.com> +Yichi Zhang Yogesh Raisinghani <46864533+raisinghanii@users.noreply.github.com> Younes Sandi Younes Sandi <65843206+Unessam@users.noreply.github.com> @@ -723,6 +753,8 @@ Yu Feng Yuji Kanagawa Yuki K Yury Kirienko +Yuvraj Pradhan +Yuvraj Pradhan Zac Hatfield-Dodds Zach Brugh <111941670+zachbrugh@users.noreply.github.com> Zé Vinícius diff --git a/doc/changelog/2.3.0-changelog.rst b/doc/changelog/2.3.0-changelog.rst new file mode 100644 index 000000000000..2f2f5f66a4df --- /dev/null +++ b/doc/changelog/2.3.0-changelog.rst @@ -0,0 +1,689 @@ + +Contributors +============ + +A total of 132 people contributed to this release. People with a "+" by their +names contributed a patch for the first time. + +* !DWesl +* !fengluoqiuwu +* !h-vetinari +* !partev +* Aarni Koskela + +* Abdu Zoghbi + +* Abhishek Kumar +* Adam J. Stewart +* Aditi Juneja +* Adrin Jalali +* Agriya Khetarpal +* Alicia Boya García + +* Andrej Zhilenkov +* Andrew Nelson +* Angus Gibson + +* Antonio Rech Santos + +* Ari Gato + +* Arnaud Tremblay + +* Arvid Bessen + +* Baskar Gopinath + +* Carlos Martin + +* Charles Harris +* Chris Navarro + +* Chris Sidebottom +* Christian Lorentzen +* Christine P. Chai + +* Christopher Sidebottom +* Clément Robert +* Colin Gilgenbach + +* Craig Peters + +* Cédric Hannotier +* Daniel Hrisca +* Derek Homeier +* Diego Baldassar + +* Dimitri Papadopoulos Orfanos +* Eoghan O'Connell + +* Eric Larson +* Ernst Peng + +* Evgeni Burovski +* Filipe Laíns +* François Rozet + +* François de Coatpont + +* GUAN MING +* Giovanni Del Monte +* Guan Ming(Wesley) Chiu + +* Gyeongjae Choi + +* Halle Loveday + +* Hannah Wheeler + +* Hao Chen + +* Harmen Stoppels + +* Hin-Tak Leung + +* Ian DesJardin + +* Ihar Hrachyshka + +* Ilhan Polat +* Inessa Pawson +* J. Steven Dodge + +* Jake VanderPlas +* Jiachen An + +* Jiuding Tan (谭九鼎) +* John Kirkham +* John Stilley + +* Jonathan Albrecht + +* Joren Hammudoglu +* Kai Germaschewski + +* Krishna Bindumadhavan + +* Lucas Colley +* Luka Krmpotić + +* Lysandros Nikolaou +* Maanas Arora +* Makima C. Yang + +* Marco Barbosa + +* Marco Edward Gorelli + +* Mark Harfouche +* Marten van Kerkwijk +* Mateusz Sokół +* Matt Haberland +* Matthew Brett +* Matthew Goldsberry + +* Matthew Sterrett +* Matthias Diener +* Matthieu Darbois +* Matti Picus +* Melissa Weber Mendonça +* Michael Siebert +* Mike O'Brien + +* Mohammed Abdul Rahman + +* Mugundan Selvanayagam + +* Musharaf Aijaz Baba + +* Musharraffaijaz + +* Nathan Goldbaum +* Nicholas Christensen + +* Nitish Satyavolu + +* Omid Rajaei +* PTUsumit + +* Peter Hawkins +* Peyton Murray +* Pieter Eendebak +* Raghuveer Devulapalli +* Ralf Gommers +* Rob Timpe + +* Robert Kern +* Rohit Goswami +* Ross Barnowski +* Roy Smart +* S.S. + +* Saransh Chopra +* Saraswathy Kalaiselvan + +* Sayed Adel +* Sebastian Berg +* Shantanu Jain + +* Shashwat Pandey + +* Shi Entong + +* Simon Altrogge +* Stan Ulbrych +* Thomas A Caswell +* Théotime Grohens + +* Tyler Reddy +* WANG Xuerui + +* WEN Hao + +* Wang Yang (杨旺) + +* Warren Weckesser +* Warrick Ball +* Wenlong + +* William Andrea +* Yakov Danishevsky + +* Yichi Zhang + +* Your Name + +* Yuvraj Pradhan + +* dependabot[bot] +* hfloveday12 + +* karl + +* samir539 + + +Pull requests merged +==================== + +A total of 543 pull requests were merged for this release. + +* `#22718 `__: DOC: Add docs on using GitHub Codespaces for NumPy development +* `#25675 `__: ENH: add matvec and vecmat gufuncs +* `#25934 `__: ENH: Convert tanh from C universal intrinsics to C++ using Highway +* `#25991 `__: ENH: Optimize polyutils as_series +* `#26018 `__: ENH add hash based unique +* `#26745 `__: ENH, DOC: Add support for interactive examples for NumPy with... +* `#26958 `__: BUG: index overlap copy +* `#27288 `__: BUG: Scalar array comparison should return np.bool +* `#27300 `__: CI: pycodestyle → ruff +* `#27309 `__: MNT: Enforce ruff/Pyflakes rules (F) +* `#27324 `__: DOC: Removing module name from by-topic docs +* `#27343 `__: ENH: Add support for flat indexing on flat iterator +* `#27404 `__: DOC: document type promotion with Python types +* `#27522 `__: ENH: Cleanup npy_find_array_wrap +* `#27523 `__: ENH: Improve performance of np.count_nonzero for float arrays +* `#27648 `__: MAINT: Fix the code style to our C-Style-Guide +* `#27738 `__: DEP: testing: disable deprecated use of keywords x/y +* `#27784 `__: BUG: ``sinc``\ : fix underflow for float16 +* `#27789 `__: ENH: Implement np.strings.slice as a gufunc +* `#27819 `__: CI: add windows free-threaded CI +* `#27823 `__: BEG, MAINT: Begin NumPy 2.3.0 development. +* `#27824 `__: BUG: Fix mismatch in definition and declaration for a couple... +* `#27826 `__: CI: update circleci to python3.11.10, limit parallel builds. +* `#27827 `__: CI: skip ninja installation in linux_qemu workflows +* `#27830 `__: ENH: speedup evaluation of numpy.polynomial.legendre.legval. +* `#27839 `__: DOC: Correct version-added for mean arg for nanvar and nanstd +* `#27841 `__: BUG: Never negate strides in reductions (for now) +* `#27847 `__: MAINT: Bump pypa/cibuildwheel from 2.21.3 to 2.22.0 +* `#27848 `__: MAINT: Bump mamba-org/setup-micromamba from 2.0.1 to 2.0.2 +* `#27850 `__: DOC: Correct versionadded for vecmat and matvec. +* `#27854 `__: MAINT: Use mask_store instead of store for compiler bug workaround +* `#27856 `__: SIMD: add lsx optimization for loongarch, and add Qemu tests +* `#27858 `__: DOC: Fix typo +* `#27860 `__: MAINT: Add helper for static or heap allocated scratch space +* `#27862 `__: MAINT: Drop Python 3.10 support. +* `#27864 `__: ENH: stack custom multiarray import exception with the original... +* `#27868 `__: BUG: fix importing numpy in Python's optimized mode +* `#27869 `__: TYP: Fix ``np.interp`` signature for scalar types +* `#27875 `__: MAINT: Ensure correct handling for very large unicode strings +* `#27877 `__: ENH: Refactor ``__qualname__`` across API +* `#27878 `__: DOC: Fix double import in docs +* `#27879 `__: DEV: Add venv files to .gitignore +* `#27883 `__: MAINT,ENH: Reorganize buffered iteration setup +* `#27884 `__: ENH: Remove unnecessary list collection +* `#27886 `__: MAINT: Move uint aligned check to actual transfer function setup +* `#27887 `__: MAINT: A few other small nditer fixes +* `#27896 `__: PERF: improve multithreaded ufunc scaling +* `#27897 `__: MAINT: Bump github/codeql-action from 3.27.5 to 3.27.6 +* `#27898 `__: MAINT: Remove ``25675.new_feature.rst`` snippet. +* `#27899 `__: TST: add timeouts for github actions tests and wheel builds. +* `#27901 `__: MAINT: simplify power fast path logic +* `#27910 `__: MAINT: Make qualname tests more specific and fix code where needed +* `#27914 `__: DOC: Remove 27896-snippet. +* `#27915 `__: MAINT: Bump actions/cache from 4.1.2 to 4.2.0 +* `#27917 `__: CI: Use hashes in specifying some actions. +* `#27920 `__: DOC: Fix invalid URL in the index.rst file. +* `#27921 `__: MAINT: Bump actions/checkout from 4.1.1 to 4.2.2 +* `#27922 `__: MAINT: Move user pointers out of axisdata and simplify iternext +* `#27923 `__: ENH: Add cython wrappers for NpyString API +* `#27927 `__: DOC: Use internal/intersphinx links for neps. +* `#27930 `__: MAINT: Fix cirrus MacOs wheel builds [wheel build] +* `#27931 `__: CI: audit with zizmor +* `#27933 `__: BUG: fix building numpy on musl s390x +* `#27936 `__: MAINT: Update main after 2.2.0 release. +* `#27940 `__: BUG: Fix potential inconsistent behaviour for high-demnsional... +* `#27943 `__: TEST: cleanups +* `#27947 `__: BUG:fix compile error libatomic link test to meson.build +* `#27955 `__: BUG: fix use-after-free error in npy_hashtable.cpp +* `#27956 `__: BLD: add missing include to fix build with freethreading +* `#27962 `__: MAINT: Bump github/codeql-action from 3.27.6 to 3.27.7 +* `#27963 `__: MAINT: Bump mamba-org/setup-micromamba from 2.0.2 to 2.0.3 +* `#27967 `__: TYP: Fix falsely rejected value types in ``ndarray.__setitem__`` +* `#27973 `__: MAINT: Apply assorted ruff/flake8-pie rules (PIE) +* `#27974 `__: MAINT: Apply ruff/flake8-implicit-str-concat rules (ISC) +* `#27975 `__: MAINT: Apply ruff/flake8-comprehensions rules (C4) +* `#27976 `__: MAINT: Apply assorted ruff/flake8-pyi rules (PYI) +* `#27978 `__: MAINT: Apply assorted ruff/flake8-simplify rules (SIM) +* `#27981 `__: DOC: Document abi3 compat +* `#27992 `__: BUG: Fix segfault in stringdtype lexsort +* `#27996 `__: MAINT: Bump github/codeql-action from 3.27.7 to 3.27.9 +* `#27997 `__: MAINT: Remove unnecessary (and not safe in free-threaded) 1-D... +* `#27998 `__: API,MAINT: Make ``NpyIter_GetTransferFlags`` public and avoid... +* `#27999 `__: DOC, MAINT: Fix typos found by codespell +* `#28001 `__: DOC: Fix documentation for np.dtype.kind to include 'T' for StringDType +* `#28003 `__: MAINT: random: Tweak module code in mtrand.pyx to fix a Cython... +* `#28004 `__: DOC: fix several doctests in dtype method docstrings +* `#28005 `__: BUG: Cython API was missing NPY_UINTP. +* `#28008 `__: BUG: Fix handling of matrix class in np.unique. +* `#28009 `__: TST: lib: Test average with object dtype and no weights. +* `#28013 `__: DOC: Fixed typos in development_advanced_debugging.rst +* `#28015 `__: MAINT: run ruff from the repository root +* `#28020 `__: CI: pin scipy-doctest to 1.5.1 +* `#28022 `__: MAINT: Add all submodules to ruff exclusion list. +* `#28023 `__: DOC: update to scipy-doctest 1.6.0 and fix tests +* `#28029 `__: MAINT: Bump actions/upload-artifact from 4.4.3 to 4.5.0 +* `#28032 `__: BUG,MAINT: Fix size bug in new alloc helper and use it in one... +* `#28033 `__: MAINT: Use userpointers to avoid NPY_MAXARGS in iternext() +* `#28035 `__: MAINT: Move ``lib.format`` and ``ctypeslib`` to submodules/private... +* `#28036 `__: Replace Twitter with X +* `#28039 `__: TYP: allow ``None`` in operand sequence of nditer +* `#28043 `__: BUG: Ensure einsum uses chunking (now that nditer doesn't) +* `#28051 `__: MAINT: Update main after 2.2.1 release. +* `#28053 `__: TYP: fix ``void`` arrays not accepting ``str`` keys in ``__setitem__`` +* `#28056 `__: BUG: Fix f2py directives and ``--lower`` casing +* `#28058 `__: MAINT: Update ``spin lint`` command +* `#28060 `__: CI: fix check that GIL remains disabled for free-threaded wheels +* `#28065 `__: TYP: fix unnecessarily broad ``integer`` binop return types +* `#28069 `__: MAINT: update NEP 29 +* `#28073 `__: TYP: use mypy_primer to surface type checking regressions +* `#28074 `__: DOC: clarify np.gradient varargs requirement for axis parameter +* `#28075 `__: MAINT: Replace usage of fixed strides with inner strides in einsum +* `#28080 `__: ENH: Allow an arbitrary number of operands in nditer +* `#28081 `__: DOC: Add release snippets for iteration changes +* `#28083 `__: MAINT: Update LICENSE Copyright to 2025 +* `#28088 `__: BUG: update check for highway compiler support +* `#28089 `__: MAINT: bump ``mypy`` to ``1.14.1`` +* `#28090 `__: DOC:Fixed docstring with example use of np.select +* `#28091 `__: MAINT: Refactor stringdtype casts.c to use cpp templates +* `#28092 `__: MAINT: LoongArch: switch away from the __loongarch64 preprocessor... +* `#28094 `__: DOC: Fix documentation example for numpy.ma.masked +* `#28100 `__: DOC: Move linalg.outer from Decompositions to Matrix and vector... +* `#28101 `__: DOC: Fix sphinx markup in source/reference/random/extending.rst +* `#28102 `__: MAINT: update oldest supported GCC version from 8.4 to 9.3 +* `#28103 `__: MAINT: random: Call np.import_array() in _examples/cython/extending_distribution... +* `#28105 `__: ENH: support no-copy pickling for any array that can be transposed... +* `#28108 `__: TYP: Better ``ndarray`` binop return types for ``float64`` &... +* `#28109 `__: TYP: Fix the incorrect ``bool`` return type of ``issubdtype`` +* `#28110 `__: TYP: Always accept ``date[time]`` in the ``datetime64`` constructor +* `#28116 `__: MAINT: random: Explicitly cast RAND_INT_MAX to double to avoid... +* `#28118 `__: BUG: Fix auxdata initialization in ufunc slow path +* `#28121 `__: MAINT: Correct NumPy 2.3 C-API versioning and version information +* `#28123 `__: BUG: move reduction initialization to ufunc initialization +* `#28127 `__: DOC: Improve slice docstrings +* `#28128 `__: BUG: Don't use C99 construct in import_array +* `#28129 `__: DEP: Deprecate ``numpy.typing.mypy_plugin`` +* `#28130 `__: CI: Fix mypy_primer comment workflow +* `#28133 `__: BUG: call PyType_Ready in f2py to avoid data races +* `#28134 `__: DEP: Deprecate ``numpy.typing.mypy_plugin``\ : The sequel +* `#28141 `__: DOC: Add instructions to build NumPy on WoA +* `#28142 `__: ENH: inline UTF-8 byte counter and make it branchless +* `#28144 `__: BUG: remove unnecessary call to PyArray_UpdateFlags +* `#28148 `__: MAINT: Replace usage of outdated fixed strides with inner strides... +* `#28149 `__: BUG: Fix ``from_float_positional`` errors for huge pads +* `#28154 `__: BUG: Avoid data race in PyArray_CheckFromAny_int +* `#28161 `__: DOC: Clarify ``np.loadtxt`` encoding argument default value in... +* `#28163 `__: MAINT: Avoid a redundant copy on ``a[...] = b`` +* `#28167 `__: DOC: fix formatting typo in basics.copies.rst +* `#28168 `__: TYP: Fix overlapping overloads issue in "2 in, 1 out" ufuncs +* `#28169 `__: TYP: preserve shape-type in ``ndarray.astype()`` +* `#28170 `__: TYP: Fix missing and spurious top-level exports +* `#28172 `__: BUG: Include Python-including headers first +* `#28179 `__: DOC: Remove duplicate wishlist tab in NEPs. +* `#28180 `__: DOC: Update links in HOWTO_RELEASE.rst +* `#28181 `__: CI: replace quansight-labs/setup-python with astral-sh/setup-uv +* `#28183 `__: MAINT: testing: specify python executable to use in extbuild +* `#28186 `__: MAINT: Update main after 2.2.2 release. +* `#28189 `__: MAINT, DOC: Add sphinx extension to allow svg images in PDF docs... +* `#28202 `__: MAINT: Use VQSORT_COMPILER_COMPATIBLE to determine if we should... +* `#28203 `__: BUG: fix data race in ``np.repeat`` +* `#28206 `__: BUG: Remove unnecessary copying and casting from out array in... +* `#28210 `__: corrected the numpy logo visibility issues on darkmode with the... +* `#28211 `__: MAINT: Hide decorator from pytest traceback +* `#28214 `__: ENH: add pkg_config entrypoint +* `#28219 `__: DOC: Add versionadded directive for axis argument in trim_zeros... +* `#28221 `__: BUG: allclose does not warn for invalid value encountered in... +* `#28222 `__: MAINT: Update highway to latest +* `#28223 `__: MAINT: Add [[maybe_unused] to silence some warnings +* `#28226 `__: DOC: Clarify ``__array__`` protocol arguments +* `#28228 `__: BUG: handle case when StringDType na_object is nan in float to... +* `#28229 `__: DOC: Fix a typo in doc/source/dev/development_workflow.rst +* `#28230 `__: DOC: FIx a link in Roadmap +* `#28231 `__: DOC: Fix external links in the navbar of neps webpage +* `#28232 `__: BUG: Fix float128 FPE handling on ARM64 with Clang compiler +* `#28234 `__: BUG: Add cpp atomic support +* `#28235 `__: MAINT: Compile fix for clang-cl on WoA +* `#28241 `__: TYP: Avoid upcasting ``float64`` in the set-ops +* `#28242 `__: CI: Fix the github label for ``TYP:`` PR's and issues +* `#28246 `__: BLD: better fix for clang / ARM compiles +* `#28250 `__: dtype.__repr__: prefer __name__ for user-defined types. +* `#28252 `__: test_casting_unittests.py: remove tuple +* `#28254 `__: MAINT: expire deprecations +* `#28258 `__: DOC: Change the scientific page link in NumPy/MATLAB +* `#28259 `__: TYP: Fix ``timedelta64.__divmod__`` and ``timedelta64.__mod__``... +* `#28262 `__: TYP: expire deprecations +* `#28263 `__: ENH: Add ARM64 (aarch64) CI testing +* `#28264 `__: DOC: Remove an invalid link in f2py-examples.rst +* `#28270 `__: TYP: Fixed missing typing information of set_printoptions +* `#28273 `__: CI: update sanitizer CI to use python compiled with ASAN and... +* `#28276 `__: BUG: fix incorrect bytes to StringDType coercion +* `#28279 `__: TYP: Fix scalar constructors +* `#28280 `__: TYP: stub ``numpy.matlib`` +* `#28281 `__: TYP: stub the missing ``numpy.testing`` modules +* `#28288 `__: DOC: Correct a typo in Intel License URL +* `#28290 `__: BUG: fix race initializing legacy dtype casts +* `#28291 `__: BUG: Prevent class-bound attr mutation in ``lib._iotools.NameValidator`` +* `#28294 `__: MAINT: Enable building tanh on vector length agnostic architectures +* `#28295 `__: TYP: stub ``numpy._globals`` +* `#28296 `__: TYP: stub ``numpy._expired_attrs_2_0`` +* `#28297 `__: TYP: stub ``numpy._configtool`` and ``numpy._distributor_init`` +* `#28298 `__: TYP: stub ``numpy.lib._iotools`` +* `#28299 `__: TYP: stub ``lib.user_array`` and ``lib._user_array_impl`` +* `#28300 `__: TYP: stub ``lib.introspect`` +* `#28301 `__: TYP: stub ``lib.recfunctions`` +* `#28302 `__: TYP: fix and improve ``numpy._core.arrayprint`` +* `#28303 `__: TYP: stub ``lib._datasource`` and fix ``lib._npyio_impl`` +* `#28304 `__: DOC: Remove reference to python2 +* `#28307 `__: MAINT: bump ``mypy`` to ``1.15.0`` +* `#28312 `__: DOC: remove references to Python 2 +* `#28319 `__: BUG: numpy.loadtxt reads only 50000 lines when skip_rows >= max_rows +* `#28320 `__: MAINT: Update actions/cache and use hash. +* `#28323 `__: DOC: Correct a typo in Exception TooHardError +* `#28327 `__: TYP: fix positional- and keyword-only params in ``astype``\ ,... +* `#28328 `__: CI: Update FreeBSD base image in ``cirrus_arm.yml`` +* `#28330 `__: ENH: Ensure ``lib._format_impl.read_array`` handles file reading... +* `#28332 `__: BUG: avoid segfault in np._core.multiarray.scalar +* `#28335 `__: MAINT: Update main after 2.2.3 release. +* `#28336 `__: DOC: Update link to Anaconda Eclipse/PyDev documentation +* `#28338 `__: MAINT: use OpenBLAS 0.3.29 +* `#28339 `__: MAIN: Update c,c++ line length to 88 +* `#28343 `__: BUG: Fix ``linalg.norm`` to handle empty matrices correctly. +* `#28350 `__: DOC: fix typo +* `#28353 `__: DOC: Make numpy.fft a clickable link to module +* `#28355 `__: BUG: safer bincount casting +* `#28358 `__: MAINT: No need to check for check for FPEs in casts to/from object +* `#28359 `__: DOC: Make the first paragraph more concise in internals.rst +* `#28361 `__: BUG: Make np.nonzero threading safe +* `#28370 `__: DOC: Revise bullet point formatting in ``arrays.promotions.rst`` +* `#28382 `__: DOC: fix C API docs for ``PyArray_Size`` +* `#28383 `__: DOC: Added links to CTypes and CFFI in Numba +* `#28386 `__: MAINT: Extend the default ruff exclude files +* `#28387 `__: DOC: fix expected exception from StringDType without string coercion +* `#28390 `__: MAINT: speed up slow test under TSAN +* `#28391 `__: CI: use free-threaded build for ASAN tests +* `#28392 `__: CI: build Linux aarch64 wheels on GitHub Actions +* `#28393 `__: BUG: Fix building on s390x with clang +* `#28396 `__: BUG: Fix crackfortran parsing error when a division occurs within... +* `#28404 `__: MAINT: remove legacy ucsnarrow module +* `#28406 `__: BUG: Include Python.h first +* `#28407 `__: BUG: sanity check ``__array_interface__`` number of dimensions +* `#28408 `__: DOC: Update link to Nix in Cross Compilation +* `#28411 `__: CI: use QEMU 9.2.2 for Linux Qemu tests +* `#28413 `__: DOC: add scimath in np.lib submodules listing +* `#28414 `__: DOC: Add missing punctuation to the random sampling page +* `#28415 `__: BLD: update cibuildwheel and build PyPy 3.11 wheels [wheel build] +* `#28421 `__: BUG: skip legacy dtype multithreaded test on 32 bit runners +* `#28426 `__: BUG: Limit the maximal number of bins for automatic histogram... +* `#28427 `__: DOC: remove mention of Poly.nickname +* `#28431 `__: MAINT: PY_VERSION_HEX simplify +* `#28436 `__: BUILD: move to manylinux_2_28 wheel builds +* `#28437 `__: DOC: fix documentation for Flag checking functions and macros +* `#28442 `__: ENH: Check for floating point exceptions in dot +* `#28444 `__: DOC: fix URL redirects +* `#28447 `__: DOC: repositioned bitwise_count under bit-wise operations +* `#28451 `__: DOC: Add -avx512_spr to disable AVX512 in build options +* `#28452 `__: TYP: stub ``random._pickle`` +* `#28453 `__: BUG: Fix return type of NpyIter_GetIterNext in Cython declarations +* `#28455 `__: MAINT: Bump mamba-org/setup-micromamba from 2.0.3 to 2.0.4 +* `#28456 `__: MAINT: Bump actions/cache from 4.2.0 to 4.2.2 +* `#28458 `__: MAINT: Bump actions/upload-artifact from 4.5.0 to 4.6.1 +* `#28459 `__: MAINT: Bump github/codeql-action from 3.27.9 to 3.28.11 +* `#28460 `__: MAINT: Bump astral-sh/setup-uv from 5.2.1 to 5.3.1 +* `#28461 `__: MAINT: Update dependabot.yml file +* `#28462 `__: TYP: Add specializations to meshgrid stubs +* `#28464 `__: MAINT: Bump actions/setup-python from 5.3.0 to 5.4.0 +* `#28465 `__: MAINT: Bump ossf/scorecard-action from 2.4.0 to 2.4.1 +* `#28466 `__: MAINT: Bump actions/checkout from 4.1.1 to 4.2.2 +* `#28467 `__: MAINT: Bump conda-incubator/setup-miniconda from 3.1.0 to 3.1.1 +* `#28468 `__: MAINT: Bump actions/download-artifact from 4.1.8 to 4.1.9 +* `#28473 `__: DOC: add typing badge to README +* `#28475 `__: MAINT: skip slow_pypy tests on pypy +* `#28477 `__: MAINT: fix typo in normal distribution functions docstrings +* `#28480 `__: ENH: Convert logical from C universal intrinsics to C++ using... +* `#28483 `__: DOC: only change tp_name on CPython +* `#28485 `__: MAINT: Bump actions/setup-python from 5.3.0 to 5.4.0 +* `#28488 `__: fix aarch64 CI run +* `#28489 `__: MAINT: Enable building loop_logical on vector length agnostic... +* `#28491 `__: TYP: fix typing errors in ``_core.shape_base`` +* `#28492 `__: TYP: fix typing errors in ``_core.strings`` +* `#28494 `__: TYP: fix typing errors in ``_core.records`` +* `#28495 `__: DOC: let docstring mention that unique_values is now unsorted +* `#28497 `__: TYP: don't use literals in shape-types +* `#28498 `__: TYP: accept non-integer shapes in array constructor without a... +* `#28499 `__: TYP: remove unneseccary cast +* `#28500 `__: TYP: stub ``numpy.random._bounded_integers`` +* `#28502 `__: TYP: stub ``numpy.random._common`` +* `#28503 `__: API: Always allow ``sorted=False`` and make a note about it +* `#28505 `__: TYP: stub ``numpy._core.umath`` +* `#28506 `__: TYP: fix typing errors in ``numpy.lib._arrayterator_impl`` +* `#28507 `__: MAINT: remove ``ma.timer_comparison`` +* `#28508 `__: TYP: fix signatures of ``ndarray.put`` and ``ndarray.view`` +* `#28509 `__: TYP: annotate the missing ``ufunc.resolve_dtypes`` method +* `#28511 `__: TYP: fix stubtest errors in ``numpy._core`` +* `#28513 `__: TYP: stub ``numpy._core.overrides`` +* `#28514 `__: TYP: stub ``numpy._utils`` +* `#28515 `__: TYP: stub ``numpy._core._dtype[_ctypes]`` +* `#28517 `__: TYP: stub the remaining ``numpy._core.\*`` modules +* `#28518 `__: TYP: stub the missing submodules of ``numpy.linalg`` +* `#28519 `__: TYP: stub ``numpy._pyinstaller`` +* `#28520 `__: TYP: stub ``numpy.fft.helper`` (deprecated) +* `#28522 `__: TYP: stub ``numpy.f2py`` +* `#28523 `__: TYP: annotate the missing deprecated ``row_stack`` function +* `#28524 `__: CI, TST: Bump to cibuildwheel 2.23 (Pyodide 0.27.0) for WASM... +* `#28525 `__: TYP: fix stubtest errors in ``numpy.dtype`` and ``numpy.dtypes.\*`` +* `#28526 `__: TYP: fix stubtest errors in ``timedelta64`` and ``object_`` +* `#28527 `__: TYP: fix stubtest errors in ``numpy.lib._function_base_impl`` +* `#28528 `__: TYP: fix stubtest errors in ``numpy.lib._arraysetops_impl`` +* `#28529 `__: TYP: fix stubtest errors in ``numpy.lib._index_tricks_impl`` +* `#28530 `__: TYP: fix stubtest errors in ``numpy.lib._twodim_base_impl`` +* `#28531 `__: ENH: Add Cygwin extensions to list to copy to CWD in f2py meson... +* `#28532 `__: DOC: minor editorial change +* `#28535 `__: TYP: fix stubtest errors in ``numpy._core`` +* `#28536 `__: TYP: fix stubtest errors in ``numpy._globals`` +* `#28537 `__: TYP: fix stubtest errors in ``numpy.mat[rix]lib`` +* `#28538 `__: TYP: fix stubtest errors in ``numpy.random`` +* `#28539 `__: TYP: fix stubtest errors in ``numpy.testing`` +* `#28540 `__: TYP: fix typing errors in ``numpy.ndarray`` +* `#28541 `__: TYP: fix stubtest error in ``numpy.ma`` +* `#28546 `__: MAINT: Update main after NumPy 2.2.4 release. +* `#28547 `__: MAINT: Bump pypa/cibuildwheel from 2.23.0 to 2.23.1 +* `#28555 `__: MAINT: Bump actions/download-artifact from 4.1.9 to 4.2.0 +* `#28556 `__: NEP 54: Change status to Accepted +* `#28560 `__: MAINT: Bump actions/download-artifact from 4.2.0 to 4.2.1 +* `#28561 `__: MAINT: Bump github/codeql-action from 3.28.11 to 3.28.12 +* `#28562 `__: MAINT: Bump actions/upload-artifact from 4.6.1 to 4.6.2 +* `#28563 `__: MAINT: Bump actions/cache from 4.2.2 to 4.2.3 +* `#28568 `__: MAINT: Bump astral-sh/setup-uv from 5.3.1 to 5.4.0 +* `#28569 `__: Fixing various spelling errors +* `#28571 `__: BLD: use ``manylinux_2_28:2025.03.23-1`` [wheel build] +* `#28576 `__: API,ENH: Allow forcing an array result in ufuncs +* `#28577 `__: BUG: avoid deadlocks with C++ shared mutex in dispatch cache +* `#28581 `__: MAINT: Bump github/codeql-action from 3.28.12 to 3.28.13 +* `#28586 `__: MAINT: Bump pypa/cibuildwheel from 2.23.1 to 2.23.2 +* `#28587 `__: MAINT: Bump actions/setup-python from 5.4.0 to 5.5.0 +* `#28591 `__: TYP: Type masked array shape, dtype, __int__, and __float__ +* `#28593 `__: TYP: Type ``numpy.ma.min`` +* `#28600 `__: BUG: Set writeable flag for writeable dlpacks. +* `#28601 `__: MAINT: Bump scientific-python/upload-nightly-action from 0.6.1... +* `#28607 `__: CI: fix cirrus config [wheel build] +* `#28611 `__: MAINT: Bump astral-sh/setup-uv from 5.4.0 to 5.4.1 +* `#28612 `__: TYP: Type ``ma.max`` and ``ma.ptp`` +* `#28615 `__: ENH: Upgrade Array API version to 2024.12 +* `#28616 `__: TYP: Type ``ma.MaskedArray.min`` +* `#28617 `__: MAINT: Bump actions/dependency-review-action from 4.5.0 to 4.6.0 +* `#28618 `__: MAINT, CI: Update Ubuntu to 22.04 in azure-pipelines +* `#28619 `__: ENH: Use openmp on x86-simd-sort to speed up np.sort and np.argsort +* `#28621 `__: DOC: Fix typo in ``numpy/typing/__init__.py`` +* `#28623 `__: TYP: Type ``ma.MaskedArray.max`` and ``ma.MaskedArray.ptp`` +* `#28624 `__: BUG: fix ``np.vectorize`` for object dtype +* `#28626 `__: DOC: update array API standard version in compatibility page +* `#28627 `__: MAINT: replace string.format() with f-strings +* `#28635 `__: BLD: Update vendor-meson to fix module_feature conflicts arguments... +* `#28637 `__: TYP: Fix overload for ``ma.MaskedArray.{min,max,ptp}`` and ``ma.{min,max,ptp}`` ... +* `#28638 `__: TYP: Type ``MaskedArray.{argmin, argmax}`` and ``np.ma.{argmin,``... +* `#28643 `__: BUG: fix deepcopying StringDType arrays +* `#28644 `__: TYP: fix ``ndarray.tolist()`` and ``.item()`` for unknown dtype +* `#28645 `__: DOC: fixes classes decorated with set_module not showing its... +* `#28647 `__: DOC: Fix typos found by codespell +* `#28649 `__: ENH: Improve np.linalg.det performance +* `#28653 `__: CI: Replace QEMU armhf with native (32-bit compatibility mode) +* `#28657 `__: TYP: simplified type-aliases in ``numpy._typing`` +* `#28660 `__: TYP: Accept objects that ``write()`` to ``str`` in ``savetxt`` +* `#28662 `__: MAINT: Remove distutils CPU dispatcher compatibility code +* `#28664 `__: TYP: Type ``MaskedArray.sort`` +* `#28666 `__: MAINT: Bump github/codeql-action from 3.28.13 to 3.28.14 +* `#28667 `__: TYP: replace ``_ScalarType`` with ``_SCT`` +* `#28668 `__: TYP: replace ``_ArrayType`` with ``_ArrayT`` +* `#28669 `__: TYP: default to ``dtype[Any]`` +* `#28671 `__: SIMD: Fix Highway QSort symbol linking error on aarch32/ASIMD +* `#28672 `__: MAINT: Bump github/codeql-action from 3.28.14 to 3.28.15 +* `#28674 `__: TYP: add missing ``"b1"`` literals for ``dtype[bool]`` +* `#28675 `__: TYP: fix and improve ``numpy.lib._type_check_impl`` +* `#28676 `__: TYP: fix mypy test failures +* `#28677 `__: TYP: Type ``MaskedArray.partition`` and ``MaskedArray.argpartition`` +* `#28678 `__: DEP: Deprecate ``.T`` property for non-2dim arrays and scalars +* `#28680 `__: TYP: Type ``MaskedArray.take`` and ``np.ma.take`` +* `#28684 `__: TYP: replace ``_DType`` with ``_DTypeT`` +* `#28688 `__: TYP: rename ``_ShapeType`` TypeVar to ``_ShapeT`` +* `#28689 `__: TYP: Type ``MaskedArray.__{ge,gt,le,lt}__`` +* `#28690 `__: TYP: replace ``_SCT`` with ``_ScalarT`` +* `#28693 `__: BLD: fix meson_version warning +* `#28695 `__: DOC: linalg.matrix_transpose: add alias note +* `#28699 `__: TYP: Fix false rejection of ``NDArray[object_].__abs__()`` +* `#28702 `__: TYP: Fix inconsistent ``NDArray[float64].__[r]truediv__`` return... +* `#28703 `__: MAINT: Improve float16 and float32 printing +* `#28710 `__: ENH: Improve performance for np.result_type +* `#28712 `__: MAINT: ``%i`` → ``%d`` +* `#28715 `__: TYP: Type ``np.ma.{is_masked,ndim,size,ids,iscontiguous}`` +* `#28717 `__: TYP: fix string-like ``ndarray`` rich comparison operators +* `#28719 `__: MAINT: switching from ``%i`` to ``fstrings`` +* `#28720 `__: TYP: drop py310 support +* `#28724 `__: STY: Apply assorted ruff rules (RUF) +* `#28725 `__: STY: Enforce ruff/pycodestyle warnings (W) +* `#28726 `__: STY: Apply assorted ruff/refurb rules (FURB) +* `#28728 `__: STY: Apply assorted ruff/pyupgrade rules (UP) +* `#28731 `__: BUG: Prevent nanmax/nanmin from copying memmap arrays +* `#28733 `__: TYP: remove ``_typing._UnknownType`` and ``_ArrayLikeUnknown`` +* `#28735 `__: TYP: Type ``MaskedArray.count`` and ``np.ma.count`` +* `#28738 `__: TYP: fix incorrect ``random.Generator.integers`` return type +* `#28739 `__: MNT: get rid of references to Python 3.10 +* `#28740 `__: MAINT: Bump astral-sh/setup-uv from 5.4.1 to 5.4.2 +* `#28741 `__: BUG: Re-enable overriding functions in the ``np.strings`` module. +* `#28742 `__: TYP: Type ``MaskedArray.filled`` and ``np.ma.filled`` +* `#28743 `__: MNT: Enforce ruff/pygrep-hooks rules (PGH) +* `#28744 `__: STY: Apply more ruff rules (RUF) +* `#28745 `__: TYP: Type ``MaskedArray.put``\ , ``np.ma.put``\ , ``np.ma.putmask`` +* `#28746 `__: TYP: ``numpy.ma`` squiggly line cleanup +* `#28747 `__: TYP: some ``[arg]partition`` fixes +* `#28748 `__: ENH: Support Python 3.14 +* `#28750 `__: TYP: fix ``count_nonzero`` signature +* `#28751 `__: MNT: discard Python 2 leftover +* `#28752 `__: MNT: Apply ruff/Pylint rule PLW0129 (assertions that never fail) +* `#28754 `__: MNT: Enforce ruff/Pylint Error rules (PLE) +* `#28755 `__: MNT: Apply assorted ruff/Pylint Refactor rules (PLR) +* `#28756 `__: MNT: Apply assorted ruff/Pylint Warning rules (PLW) +* `#28757 `__: BUG: Fix AVX512_SPR dispatching for SVML half-precision operations +* `#28760 `__: STY: Apply ruff/pyupgrade rule UP032 +* `#28763 `__: STY: Use f-string instead of ``format`` call +* `#28764 `__: MNT: Enforce ruff rules: Flynt (FLY) and flake8-pie (PIE) +* `#28765 `__: MNT: Enforce ruff/flake8-bugbear rules (B) +* `#28766 `__: TYP: Type ``MaskedArray.compressed`` and ``np.ma.compressed`` +* `#28768 `__: MAINT: getting rid of old ``%`` and ``.format(...)`` strings... +* `#28769 `__: ENH: Improve Floating Point Cast Performance on ARM +* `#28770 `__: MNT: Enforce ruff/pyupgrade rules (UP) +* `#28771 `__: ENH: Include offset in error message when fallocate() fails +* `#28775 `__: STY: Partially apply ruff/pycodestyle rules (E) +* `#28779 `__: MAINT: Update main after Numpy 2.2.5 release +* `#28789 `__: BUG: Re-enable GCC function-specific optimization attributes +* `#28793 `__: TYP: Type ``np.ma.allclose`` and ``np.ma.allequal`` +* `#28798 `__: TST: skip test if spawning threads triggers a RuntimeError +* `#28803 `__: MAINT: Bump github/codeql-action from 3.28.15 to 3.28.16 +* `#28804 `__: BUG: fix heap buffer overflow in np.strings.find +* `#28806 `__: BUG: Fix `` __array__(None)`` to preserve dtype +* `#28807 `__: TYP: fix ``NDArray[floating] + float`` return type +* `#28808 `__: CI: Make clang_TSAN CI job use cpython_sanity docker image +* `#28809 `__: TYP: write ``dtype[Any]`` as ``dtype`` +* `#28810 `__: TYP: replace ``_Self`` type parameters with ``typing.Self`` +* `#28811 `__: TYP: remove unnecessary scalar-type ``Any`` type-args +* `#28816 `__: MAINT: Bump actions/setup-python from 5.5.0 to 5.6.0 +* `#28817 `__: MAINT: Bump astral-sh/setup-uv from 5.4.2 to 6.0.0 +* `#28818 `__: MAINT: Bump actions/download-artifact from 4.2.1 to 4.3.0 +* `#28819 `__: TYP: simplify redundant unions of builtin scalar types +* `#28820 `__: TYP: ``None`` at the end of a union +* `#28821 `__: BUG: Use unrotated companion matrix in polynomial.polyroots. +* `#28831 `__: TYP: Fix type annotations for ``np.ma.nomask`` and ``np.ma.MaskType`` +* `#28832 `__: TYP: Type ``np.ma.getmask`` +* `#28833 `__: TYP: Type ``np.ma.is_mask`` +* `#28836 `__: ENH: Provide Windows 11 ARM64 wheels (#22530) +* `#28841 `__: BUG: Fix Clang warning in loops_half.dispatch.c.src +* `#28845 `__: TYP: Type ``MaskedArray.nonzero`` +* `#28847 `__: TYP: Use _Array1D alias in ``numpy.ma.core.pyi`` +* `#28848 `__: TYP: Type ``MaskedArray.ravel`` +* `#28849 `__: TYP: Type ``MaskedArray.repeat``\ , improve overloads for ``NDArray.repeat``\... +* `#28850 `__: TYP: Type ``MaskedArray.swapaxes`` +* `#28854 `__: MAINT: Bump pypa/cibuildwheel from 2.23.2 to 2.23.3 +* `#28855 `__: TYP: add missing ``mod`` params to ``__[r]pow__`` +* `#28856 `__: TYP: generic ``StringDType`` +* `#28857 `__: TYP: implicit ``linalg`` private submodule re-exports +* `#28858 `__: TYP: fix the ``set_module`` signature +* `#28859 `__: DOC: Replace http:// with https:// +* `#28860 `__: BLD: update vendored Meson: v1.6.1 and iOS support +* `#28862 `__: BUG: fix stringdtype singleton thread safety +* `#28863 `__: TYP: Improve consistency of (masked) array typing aliases +* `#28867 `__: TYP: Type ``MaskedArray.{__setmask__,mask,harden_mask,soften_mask,hardmask,unsha``... +* `#28868 `__: TYP: Type ``MaskedArray.{imag, real, baseclass, mT}`` +* `#28869 `__: MAINT: Bump astral-sh/setup-uv from 6.0.0 to 6.0.1 +* `#28870 `__: MNT: retire old script for SVN repositories +* `#28871 `__: MNT: retire script superseded by ruff rule W605 +* `#28872 `__: DOC: consistent and updated LICENSE files for wheels +* `#28874 `__: DOC: ``numpy.i`` will not be included as part of SWIG +* `#28876 `__: MNT: discard unused function using os.system() +* `#28877 `__: DOC: update content of cross compilation build docs +* `#28878 `__: STY: Enforce more ruff rules +* `#28879 `__: STY: Apply assorted ruff/refurb rules (FURB) +* `#28880 `__: TYP: Type ``MaskedArray.all`` and ``MaskedArray.any`` +* `#28882 `__: MAINT: address warning in SWIG tests +* `#28883 `__: MAINT: from_dlpack thread safety fixes +* `#28884 `__: DEP: deprecate ``numpy.typing.NBitBase`` +* `#28887 `__: MAINT: Bump github/codeql-action from 3.28.16 to 3.28.17 +* `#28888 `__: DOC: math mode x to \times in docstring for numpy.linalg.multi_dot +* `#28892 `__: MAINT: Avoid dereferencing/strict aliasing warnings +* `#28893 `__: TYP: remove non-existent extended-precision scalar types +* `#28898 `__: BUG: Fix missing check for PyErr_Occurred() in _pyarray_correlate. +* `#28904 `__: BLD: update vendored Meson to include iOS fix +* `#28905 `__: TYP: Test ``MaskedArray.transpose`` and ``MaskedArray.T``\ ,... +* `#28906 `__: TYP: np.argmin and np.argmax overload changes +* `#28908 `__: TYP: Fix non-existent ``CanIndex`` annotation in ``ndarray.setfield`` +* `#28912 `__: TYP: add ``float64`` overloads to ``{lin,log,geom}space`` +* `#28918 `__: DOC: Fixes absent line numbers on link to classes decorated with... +* `#28923 `__: BUG: Use string conversion defined on dtype for .str +* `#28927 `__: MAINT: Remove outdated ``MaskedArray.__div__`` and ``MaskedArray.__idiv__`` +* `#28928 `__: MNT: add support for 3.14.0b1 +* `#28929 `__: MAINT: remove py2 ``__div__`` methods from ``poly1d`` and ``ABCPolyBase`` +* `#28930 `__: MAINT: remove py2 ``__div__`` remnants from the tests +* `#28931 `__: MAINT: remove py2 ``__div__`` methods from ``lib.user_array.container`` +* `#28932 `__: MAINT: remove references to 256-bits extended precision types +* `#28933 `__: MAINT: Use consistent naming for ``numpy/typing/tests/data/fail/ma.pyi`` +* `#28934 `__: TYP, TST: improved type-testing +* `#28935 `__: MAINT: Enable ruff E251 +* `#28936 `__: TST: Prevent import error when tests are not included in the... +* `#28937 `__: CI: fix TSAN CI by using a different docker image +* `#28938 `__: MNT: clean up free-threaded CI configuration +* `#28939 `__: MAINT: Bump actions/dependency-review-action from 4.6.0 to 4.7.0 +* `#28940 `__: TYP: optional type parameters for ``ndarray`` and ``flatiter`` +* `#28941 `__: DOC: Fix titles in ``development_ghcodespaces.rst`` +* `#28945 `__: MAINT: Enable linting with ruff E501 +* `#28952 `__: MAINT: Bump actions/dependency-review-action from 4.7.0 to 4.7.1 +* `#28954 `__: MAINT: Enable linting with ruff E501 for numpy._core +* `#28956 `__: DOC: Remove references to Python 2/3 +* `#28958 `__: TYP: reject complex scalar types in ``ndarray.__ifloordiv__`` +* `#28959 `__: TYP: remove redundant ``ndarray`` inplace operator overloads +* `#28960 `__: TYP: fix mypy & pyright errors in ``np.matrix`` +* `#28961 `__: DEP: finalize removal of ``numpy.compat`` +* `#28962 `__: TYP: type-testing without the mypy plugin +* `#28963 `__: MAINT: Update ruff to 0.11.9 in linting requirements +* `#28969 `__: MNT: Enforce ruff/isort rules (I) +* `#28971 `__: MAINT: Enable linting with ruff E501 +* `#28972 `__: MNT: Get rif of ``# pylint: `` pragma controls +* `#28974 `__: MNT: Get rid of ``version: $Id`` CVS tags +* `#28975 `__: MNT: import numpy as np +* `#28976 `__: MNT: Get rid of Pyflakes / flake8 +* `#28977 `__: MNT: Enforce ruff/flake8-implicit-str-concat rules (ISC) +* `#28978 `__: MNT: Enforce ruff/pandas-vet rules (PD) +* `#28981 `__: STY: reformat the ``_typing`` imports without trailing commas +* `#28982 `__: TYP: Gradual shape type defaults +* `#28984 `__: MNT: Use isinstance() instead of comparing type() +* `#28986 `__: TYP: Type ``MaskedArray.__{iadd,isub,imul,itruediv,ifloordiv,ipow}__`` +* `#28987 `__: MNT: Align ruff pin between ``requirements/linter_requirements.txt``... +* `#28988 `__: TYP: add missing ``ndarray.__{add,mul}__`` ``character`` type... +* `#28989 `__: MAINT: Bump github/codeql-action from 3.28.17 to 3.28.18 +* `#28990 `__: Revert "DEP: Deprecate ``.T`` property for non-2dim arrays and... +* `#28993 `__: MAINT: update NPY_FEATURE_VERSION after dropping python 3.10 +* `#28994 `__: TYP: allow inplace division of ``NDArray[timedelta64]`` by floats +* `#28995 `__: TYP: remove ``from __future__ import annotations`` +* `#28998 `__: MAINT: Update main after 2.2.6 release. +* `#29002 `__: MAINT: Update download-wheels for multiple pages +* `#29006 `__: ENH: Disable the alloc cache under address and memory sanitizers +* `#29008 `__: MNT: fix CI issues on main +* `#29018 `__: MAINT: Bump larsoner/circleci-artifacts-redirector-action from... +* `#29033 `__: BUG: Fix workflow error +* `#29042 `__: MNT: constant string arrays instead of pointers in C +* `#29043 `__: BUG: Avoid compile errors in f2py modules +* `#29044 `__: BUG: Fix f2py derived types in modules +* `#29046 `__: BUG: Fix cache use regression + diff --git a/doc/release/upcoming_changes/26018.change.rst b/doc/release/upcoming_changes/26018.change.rst deleted file mode 100644 index 9d7c139be183..000000000000 --- a/doc/release/upcoming_changes/26018.change.rst +++ /dev/null @@ -1,7 +0,0 @@ -``unique_values`` may return unsorted data ------------------------------------------- -The relatively new function (added in NumPy 2.0) ``unique_values`` may now -return unsorted results. Just as ``unique_counts`` and ``unique_all`` -these never guaranteed a sorted result, however, the result -was sorted until now. In cases where these do return a sorted result, this -may change in future releases to improve performance. diff --git a/doc/release/upcoming_changes/26018.performance.rst b/doc/release/upcoming_changes/26018.performance.rst deleted file mode 100644 index ffeab51dbdf6..000000000000 --- a/doc/release/upcoming_changes/26018.performance.rst +++ /dev/null @@ -1,7 +0,0 @@ -Performance improvements to ``np.unique`` ------------------------------------------ -``np.unique`` now tries to use a hash table to find unique values instead of sorting -values before finding unique values. This is limited to certain dtypes for now, and -the function is now faster for those dtypes. The function now also exposes a ``sorted`` -parameter to allow returning unique values as they were found, instead of sorting them -afterwards. \ No newline at end of file diff --git a/doc/release/upcoming_changes/26745.highlight.rst b/doc/release/upcoming_changes/26745.highlight.rst deleted file mode 100644 index 5636f919c80d..000000000000 --- a/doc/release/upcoming_changes/26745.highlight.rst +++ /dev/null @@ -1,10 +0,0 @@ -Interactive examples in the NumPy documentation ------------------------------------------------ - -The NumPy documentation includes a number of examples that -can now be run interactively in your browser using WebAssembly -and Pyodide. - -Please note that the examples are currently experimental in -nature and may not work as expected for all methods in the -public API. diff --git a/doc/release/upcoming_changes/27288.improvement.rst b/doc/release/upcoming_changes/27288.improvement.rst deleted file mode 100644 index c7319554c63f..000000000000 --- a/doc/release/upcoming_changes/27288.improvement.rst +++ /dev/null @@ -1,3 +0,0 @@ -* Scalar comparisons between non-comparable dtypes such as - `np.array(1) == np.array('s')` now return a NumPy bool instead of - a Python bool. \ No newline at end of file diff --git a/doc/release/upcoming_changes/27789.new_function.rst b/doc/release/upcoming_changes/27789.new_function.rst deleted file mode 100644 index 734a0c3bc2b5..000000000000 --- a/doc/release/upcoming_changes/27789.new_function.rst +++ /dev/null @@ -1,5 +0,0 @@ -New function `numpy.strings.slice` ----------------------------------- -The new function `numpy.strings.slice` was added, which implements fast -native slicing of string arrays. It supports the full slicing API including -negative slice offsets and steps. \ No newline at end of file diff --git a/doc/release/upcoming_changes/27883.c_api.rst b/doc/release/upcoming_changes/27883.c_api.rst deleted file mode 100644 index 107e0036c5c2..000000000000 --- a/doc/release/upcoming_changes/27883.c_api.rst +++ /dev/null @@ -1,4 +0,0 @@ -* `NpyIter_GetTransferFlags` is now available to check if - the iterator needs the Python API or if casts may cause floating point - errors (FPE). FPEs can for example be set when casting ``float64(1e300)`` - to ``float32`` (overflow to infinity) or a NaN to an integer (invalid value). \ No newline at end of file diff --git a/doc/release/upcoming_changes/27883.change.rst b/doc/release/upcoming_changes/27883.change.rst deleted file mode 100644 index ea68771efba3..000000000000 --- a/doc/release/upcoming_changes/27883.change.rst +++ /dev/null @@ -1,17 +0,0 @@ -Changes to the main iterator and potential numerical changes ------------------------------------------------------------- -The main iterator, used in math functions and via ``np.nditer`` from Python -and ``NpyIter`` in C, now behaves differently for some buffered iterations. -This means that: - -* The buffer size used will often be smaller than the maximum buffer sized - allowed by the ``buffersize`` parameter. -* The "growinner" flag is now honored with buffered reductions when no operand - requires buffering. - -For ``np.sum()`` such changes in buffersize may slightly change numerical -results of floating point operations. -Users who use "growinner" for custom reductions could notice -changes in precision (for example, in NumPy we removed it from -``einsum`` to avoid most precision changes and improve precision -for some 64bit floating point inputs). diff --git a/doc/release/upcoming_changes/27998.c_api.rst b/doc/release/upcoming_changes/27998.c_api.rst deleted file mode 100644 index edc6371af1f9..000000000000 --- a/doc/release/upcoming_changes/27998.c_api.rst +++ /dev/null @@ -1,10 +0,0 @@ -New `NpyIter_GetTransferFlags` and ``NpyIter_IterationNeedsAPI`` change ------------------------------------------------------------------------ -NumPy now has the new `NpyIter_GetTransferFlags` function as a more precise -way checking of iterator/buffering needs. I.e. whether the Python API/GIL is -required or floating point errors may occur. -This function is also faster if you already know your needs without buffering. - -The ``NpyIter_IterationNeedsAPI`` function now performs all the checks that were -previously performed at setup time. While it was never necessary to call it -multiple times, doing so will now have a larger cost. diff --git a/doc/release/upcoming_changes/28080.c_api.rst b/doc/release/upcoming_changes/28080.c_api.rst deleted file mode 100644 index f72be7ef52fe..000000000000 --- a/doc/release/upcoming_changes/28080.c_api.rst +++ /dev/null @@ -1 +0,0 @@ -* ``NpyIter`` now has no limit on the number of operands it supports. diff --git a/doc/release/upcoming_changes/28080.improvement.rst b/doc/release/upcoming_changes/28080.improvement.rst deleted file mode 100644 index 19b85ae3c96a..000000000000 --- a/doc/release/upcoming_changes/28080.improvement.rst +++ /dev/null @@ -1,2 +0,0 @@ -* ``np.nditer`` now has no limit on the number of supported operands - (C-integer). diff --git a/doc/release/upcoming_changes/28102.change.rst b/doc/release/upcoming_changes/28102.change.rst deleted file mode 100644 index bd54378a652e..000000000000 --- a/doc/release/upcoming_changes/28102.change.rst +++ /dev/null @@ -1,6 +0,0 @@ -The minimum supported GCC version is now 9.3.0 ----------------------------------------------- -The minimum supported version was updated from 8.4.0 to 9.3.0, -primarily in order to reduce the chance of platform-specific bugs in old GCC -versions from causing issues. - diff --git a/doc/release/upcoming_changes/28105.improvement.rst b/doc/release/upcoming_changes/28105.improvement.rst deleted file mode 100644 index 537467575234..000000000000 --- a/doc/release/upcoming_changes/28105.improvement.rst +++ /dev/null @@ -1,2 +0,0 @@ -* No-copy pickling is now supported for any - array that can be transposed to a C-contiguous array. \ No newline at end of file diff --git a/doc/release/upcoming_changes/28129.deprecation.rst b/doc/release/upcoming_changes/28129.deprecation.rst deleted file mode 100644 index b1beb0c5cca3..000000000000 --- a/doc/release/upcoming_changes/28129.deprecation.rst +++ /dev/null @@ -1,4 +0,0 @@ -* The ``numpy.typing.mypy_plugin`` has been deprecated in favor of platform-agnostic - static type inference. Please remove ``numpy.typing.mypy_plugin`` from the ``plugins`` - section of your mypy configuration. If this change results in new errors being - reported, kindly open an issue. diff --git a/doc/release/upcoming_changes/28205.improvement.rst b/doc/release/upcoming_changes/28205.improvement.rst deleted file mode 100644 index 42eaaac98363..000000000000 --- a/doc/release/upcoming_changes/28205.improvement.rst +++ /dev/null @@ -1,6 +0,0 @@ -Added warnings to `np.isclose` ---------------------------------- -Added warning messages if at least one of atol or rtol are -either `np.nan` or `np.inf` within `np.isclose` - -* Warnings follow the user's `np.seterr` settings diff --git a/doc/release/upcoming_changes/28214.new_feature.rst b/doc/release/upcoming_changes/28214.new_feature.rst deleted file mode 100644 index eb95a0739e79..000000000000 --- a/doc/release/upcoming_changes/28214.new_feature.rst +++ /dev/null @@ -1,23 +0,0 @@ -NumPy now registers its pkg-config paths with the pkgconf_ PyPI package ------------------------------------------------------------------------ - -The pkgconf_ PyPI package provides an interface for projects like NumPy to -register their own paths to be added to the pkg-config search path. This means -that when using pkgconf_ from PyPI, NumPy will be discoverable without needing -for any custom environment configuration. - -.. attention:: Attention - - This only applies when using the pkgconf_ package from PyPI_, or put another - way, this only applies when installing pkgconf_ via a Python package - manager. - - If you are using ``pkg-config`` or ``pkgconf`` provided by your system, or - any other source that does not use the pkgconf-pypi_ project, the NumPy - pkg-config directory will not be automatically added to the search path. In - these situations, you might want to use ``numpy-config``. - - -.. _pkgconf: https://github.com/pypackaging-native/pkgconf-pypi -.. _PyPI: https://pypi.org/ -.. _pkgconf-pypi: https://github.com/pypackaging-native/pkgconf-pypi diff --git a/doc/release/upcoming_changes/28250.improvement.rst b/doc/release/upcoming_changes/28250.improvement.rst deleted file mode 100644 index 703a8bb0c2e1..000000000000 --- a/doc/release/upcoming_changes/28250.improvement.rst +++ /dev/null @@ -1,3 +0,0 @@ -* The ``__repr__`` for user-defined dtypes now prefers the ``__name__`` of the - custom dtype over a more generic name constructed from its ``kind`` and - ``itemsize``. diff --git a/doc/release/upcoming_changes/28254.expired.rst b/doc/release/upcoming_changes/28254.expired.rst deleted file mode 100644 index 5f391eb6cbe2..000000000000 --- a/doc/release/upcoming_changes/28254.expired.rst +++ /dev/null @@ -1,29 +0,0 @@ -* Remove deprecated macros like ``NPY_OWNDATA`` from cython interfaces in favor of ``NPY_ARRAY_OWNDATA`` (deprecated since 1.7) -* Remove ``numpy/npy_1_7_deprecated_api.h`` and C macros like ``NPY_OWNDATA`` in favor of ``NPY_ARRAY_OWNDATA`` (deprecated since 1.7) -* Remove alias ``generate_divbyzero_error`` to ``npy_set_floatstatus_divbyzero`` and ``generate_overflow_error`` to ``npy_set_floatstatus_overflow`` (deprecated since 1.10) -* Remove ``np.tostring`` (deprecated since 1.19) -* Raise on ``np.conjugate`` of non-numeric types (deprecated since 1.13) -* Raise when using ``np.bincount(...minlength=None)``, use 0 instead (deprecated since 1.14) -* Passing ``shape=None`` to functions with a non-optional shape argument errors, use ``()`` instead (deprecated since 1.20) -* Inexact matches for ``mode`` and ``searchside`` raise (deprecated since 1.20) -* Setting ``__array_finalize__ = None`` errors (deprecated since 1.23) -* ``np.fromfile`` and ``np.fromstring`` error on bad data, previously they would guess (deprecated since 1.18) -* ``datetime64`` and ``timedelta64`` construction with a tuple no longer accepts an ``event`` value, either use a two-tuple of (unit, num) or a 4-tuple of (unit, num, den, 1) (deprecated since 1.14) -* When constructing a ``dtype`` from a class with a ``dtype`` attribute, that attribute must be a dtype-instance rather than a thing that can be parsed as a dtype instance (deprecated in 1.19). At some point the whole construct of using a dtype attribute will be deprecated (see #25306) -* Passing booleans as partition index errors (deprecated since 1.23) -* Out-of-bounds indexes error even on empty arrays (deprecated since 1.20) -* ``np.tostring`` has been removed, use ``tobytes`` instead (deprecated since 1.19) -* Disallow make a non-writeable array writeable for arrays with a base that do not own their data (deprecated since 1.17) -* ``concatenate()`` with ``axis=None`` uses ``same-kind`` casting by default, not ``unsafe`` (deprecated since 1.20) -* Unpickling a scalar with object dtype errors (deprecated since 1.20) -* The binary mode of ``fromstring`` now errors, use ``frombuffer`` instead (deprecated since 1.14) -* Converting ``np.inexact`` or ``np.floating`` to a dtype errors (deprecated since 1.19) -* Converting ``np.complex``, ``np.integer``, ``np.signedinteger``, ``np.unsignedinteger``, ``np.generic`` to a dtype errors (deprecated since 1.19) -* The Python built-in ``round`` errors for complex scalars. Use ``np.round`` or ``scalar.round`` instead (deprecated since 1.19) -* 'np.bool' scalars can no longer be interpreted as an index (deprecated since 1.19) -* Parsing an integer via a float string is no longer supported. (deprecated since 1.23) To avoid this error you can - * make sure the original data is stored as integers. - * use the ``converters=float`` keyword argument. - * Use ``np.loadtxt(...).astype(np.int64)`` -* The use of a length 1 tuple for the ufunc ``signature`` errors. Use ``dtype`` or fill the tuple with ``None`` (deprecated since 1.19) -* Special handling of matrix is in np.outer is removed. Convert to a ndarray via ``matrix.A`` (deprecated since 1.20) diff --git a/doc/release/upcoming_changes/28343.change.rst b/doc/release/upcoming_changes/28343.change.rst deleted file mode 100644 index 378ef775b62e..000000000000 --- a/doc/release/upcoming_changes/28343.change.rst +++ /dev/null @@ -1 +0,0 @@ -* The vector norm ``ord=inf`` and the matrix norms ``ord={1, 2, inf, 'nuc'}`` now always returns zero for empty arrays. Empty arrays have at least one axis of size zero. This affects `np.linalg.norm`, `np.linalg.vector_norm`, and `np.linalg.matrix_norm`. Previously, NumPy would raises errors or return zero depending on the shape of the array. diff --git a/doc/release/upcoming_changes/28426.change.rst b/doc/release/upcoming_changes/28426.change.rst deleted file mode 100644 index d1c48640eed0..000000000000 --- a/doc/release/upcoming_changes/28426.change.rst +++ /dev/null @@ -1,6 +0,0 @@ -Changes to automatic bin selection in numpy.histogram ------------------------------------------------------ -The automatic bin selection algorithm in ``numpy.histogram`` has been modified -to avoid out-of-memory errors for samples with low variation. -For full control over the selected bins the user can use set -the ``bin`` or ``range`` parameters of ``numpy.histogram``. diff --git a/doc/release/upcoming_changes/28436.change.rst b/doc/release/upcoming_changes/28436.change.rst deleted file mode 100644 index 60149e55a4d0..000000000000 --- a/doc/release/upcoming_changes/28436.change.rst +++ /dev/null @@ -1,10 +0,0 @@ -Build manylinux_2_28 wheels ---------------------------- - -Wheels for linux systems will use the ``manylinux_2_28`` tag (instead of the ``manylinux2014`` tag), which means -dropping support for redhat7/centos7, amazonlinux2, debian9, ubuntu18.04, and -other pre-glibc2.28 operating system versions, as per the `PEP 600 support -table`_. - -.. _`PEP 600 support table`: https://github.com/mayeut/pep600_compliance?tab=readme-ov-file#pep600-compliance-check - diff --git a/doc/release/upcoming_changes/28442.improvement.rst b/doc/release/upcoming_changes/28442.improvement.rst deleted file mode 100644 index 16d71bde19c5..000000000000 --- a/doc/release/upcoming_changes/28442.improvement.rst +++ /dev/null @@ -1 +0,0 @@ -* ``np.dot`` now reports floating point exceptions. diff --git a/doc/release/upcoming_changes/28569.change.rst b/doc/release/upcoming_changes/28569.change.rst deleted file mode 100644 index f9d26fda0484..000000000000 --- a/doc/release/upcoming_changes/28569.change.rst +++ /dev/null @@ -1,2 +0,0 @@ -* A spelling error in the error message returned when converting a string to a float with the - method ``np.format_float_positional`` has been fixed. diff --git a/doc/release/upcoming_changes/28576.new_feature.rst b/doc/release/upcoming_changes/28576.new_feature.rst deleted file mode 100644 index 2c50887a49f2..000000000000 --- a/doc/release/upcoming_changes/28576.new_feature.rst +++ /dev/null @@ -1,15 +0,0 @@ -Allow ``out=...`` in ufuncs to ensure array result --------------------------------------------------- -NumPy has the sometimes difficult behavior that it currently usually -returns scalars rather than 0-D arrays (even if the inputs were 0-D arrays). -This is especially problematic for non-numerical dtypes (e.g. ``object``). - -For ufuncs (i.e. most simple math functions) it is now possible -to use ``out=...`` (literally `...`, e.g. ``out=Ellipsis``) which is identical in behavior to ``out`` not -being passed, but will ensure a non-scalar return. -This spelling is borrowed from ``arr1d[0, ...]`` where the ``...`` -also ensures a non-scalar return. - -Other functions with an ``out=`` kwarg should gain support eventually. -Downstream libraries that interoperate via ``__array_ufunc__`` or -``__array_function__`` may need to adapt to support this. diff --git a/doc/release/upcoming_changes/28615.change.rst b/doc/release/upcoming_changes/28615.change.rst deleted file mode 100644 index 58b751e40704..000000000000 --- a/doc/release/upcoming_changes/28615.change.rst +++ /dev/null @@ -1,5 +0,0 @@ -* NumPy's ``__array_api_version__`` was upgraded from ``2023.12`` to ``2024.12``. -* `numpy.count_nonzero` for ``axis=None`` (default) now returns a NumPy scalar - instead of a Python integer. -* The parameter ``axis`` in `numpy.take_along_axis` function has now a default - value of ``-1``. diff --git a/doc/release/upcoming_changes/28619.highlight.rst b/doc/release/upcoming_changes/28619.highlight.rst deleted file mode 100644 index 6c296b92899e..000000000000 --- a/doc/release/upcoming_changes/28619.highlight.rst +++ /dev/null @@ -1,6 +0,0 @@ -Building NumPy with OpenMP Parallelization -------------------------------------------- -NumPy now supports OpenMP parallel processing capabilities when built with the -``-Denable_openmp=true`` Meson build flag. This feature is disabled by default. -When enabled, ``np.sort`` and ``np.argsort`` functions can utilize OpenMP for -parallel thread execution, improving performance for these operations. diff --git a/doc/release/upcoming_changes/28619.performance.rst b/doc/release/upcoming_changes/28619.performance.rst deleted file mode 100644 index 904decbe0ba6..000000000000 --- a/doc/release/upcoming_changes/28619.performance.rst +++ /dev/null @@ -1,7 +0,0 @@ -Performance improvements to ``np.sort`` and ``np.argsort`` ----------------------------------------------------------- -``np.sort`` and ``np.argsort`` functions now can leverage OpenMP for parallel -thread execution, resulting in up to 3.5x speedups on x86 architectures with -AVX2 or AVX-512 instructions. This opt-in feature requires NumPy to be built -with the -Denable_openmp Meson flag. Users can control the number of threads -used by setting the OMP_NUM_THREADS environment variable. diff --git a/doc/release/upcoming_changes/28669.new_feature.rst b/doc/release/upcoming_changes/28669.new_feature.rst deleted file mode 100644 index 2953a5123ccc..000000000000 --- a/doc/release/upcoming_changes/28669.new_feature.rst +++ /dev/null @@ -1,3 +0,0 @@ -* The type parameter of ``np.dtype`` now defaults to ``typing.Any``. - This way, static type-checkers will infer ``dtype: np.dtype`` as - ``dtype: np.dtype[Any]``, without reporting an error. diff --git a/doc/release/upcoming_changes/28703.change.rst b/doc/release/upcoming_changes/28703.change.rst deleted file mode 100644 index 87bb431951f9..000000000000 --- a/doc/release/upcoming_changes/28703.change.rst +++ /dev/null @@ -1,3 +0,0 @@ -* Printing of ``np.float16`` and ``np.float32`` scalars and arrays have been improved by - adjusting the transition to scientific notation based on the floating point precision. - A new legacy ``np.printoptions`` mode ``'2.2'`` has been added for backwards compatibility. diff --git a/doc/release/upcoming_changes/28713.change.rst b/doc/release/upcoming_changes/28713.change.rst deleted file mode 100644 index 5e5c5adde88b..000000000000 --- a/doc/release/upcoming_changes/28713.change.rst +++ /dev/null @@ -1 +0,0 @@ -Remove use of -Wl,-ld_classic on macOS. This hack is no longer needed by Spack, and results in libraries that cannot link to other libraries built with ld (new). diff --git a/doc/release/upcoming_changes/28741.change.rst b/doc/release/upcoming_changes/28741.change.rst deleted file mode 100644 index ca9531f490d8..000000000000 --- a/doc/release/upcoming_changes/28741.change.rst +++ /dev/null @@ -1 +0,0 @@ -Re-enable overriding functions in the :mod:`numpy.strings` module. diff --git a/doc/release/upcoming_changes/28769.performance.rst b/doc/release/upcoming_changes/28769.performance.rst deleted file mode 100644 index 7fb8f02282f6..000000000000 --- a/doc/release/upcoming_changes/28769.performance.rst +++ /dev/null @@ -1,8 +0,0 @@ -Performance improvements for ``np.float16`` casts --------------------------------------------------- -Earlier, floating point casts to and from ``np.float16`` types -were emulated in software on all platforms. - -Now, on ARM devices that support Neon float16 intrinsics (such as -recent Apple Silicon), the native float16 path is used to achieve -the best performance. diff --git a/doc/release/upcoming_changes/28856.improvement.rst b/doc/release/upcoming_changes/28856.improvement.rst deleted file mode 100644 index 83911035f097..000000000000 --- a/doc/release/upcoming_changes/28856.improvement.rst +++ /dev/null @@ -1,5 +0,0 @@ -* ``np.dtypes.StringDType`` is now a - `generic type `_ which - accepts a type argument for ``na_object`` that defaults to ``typing.Never``. - For example, ``StringDType(na_object=None)`` returns a ``StringDType[None]``, - and ``StringDType()`` returns a ``StringDType[typing.Never]``. diff --git a/doc/release/upcoming_changes/28884.deprecation.rst b/doc/release/upcoming_changes/28884.deprecation.rst deleted file mode 100644 index c1be55fb0dd3..000000000000 --- a/doc/release/upcoming_changes/28884.deprecation.rst +++ /dev/null @@ -1,28 +0,0 @@ -``numpy.typing.NBitBase`` deprecation -------------------------------------- -The ``numpy.typing.NBitBase`` type has been deprecated and will be removed in a future version. - -This type was previously intended to be used as a generic upper bound for type-parameters, for example: - -.. code-block:: python - - import numpy as np - import numpy.typing as npt - - def f[NT: npt.NBitBase](x: np.complexfloating[NT]) -> np.floating[NT]: ... - -But in NumPy 2.2.0, ``float64`` and ``complex128`` were changed to concrete subtypes, causing static type-checkers to reject ``x: np.float64 = f(np.complex128(42j))``. - -So instead, the better approach is to use ``typing.overload``: - -.. code-block:: python - - import numpy as np - from typing import overload - - @overload - def f(x: np.complex64) -> np.float32: ... - @overload - def f(x: np.complex128) -> np.float64: ... - @overload - def f(x: np.clongdouble) -> np.longdouble: ... diff --git a/doc/release/upcoming_changes/28940.new_feature.rst b/doc/release/upcoming_changes/28940.new_feature.rst deleted file mode 100644 index e0d3dc8888c3..000000000000 --- a/doc/release/upcoming_changes/28940.new_feature.rst +++ /dev/null @@ -1,6 +0,0 @@ -* Static type-checkers now interpret: - - - ``_: np.ndarray`` as ``_: npt.NDArray[typing.Any]``. - - ``_: np.flatiter`` as ``_: np.flatiter[np.ndarray]``. - - This is because their type parameters now have default values. diff --git a/doc/release/upcoming_changes/28961.expired.rst b/doc/release/upcoming_changes/28961.expired.rst deleted file mode 100644 index 92031de35e62..000000000000 --- a/doc/release/upcoming_changes/28961.expired.rst +++ /dev/null @@ -1 +0,0 @@ -* Removed the ``np.compat`` package source code (removed in 2.0) diff --git a/doc/source/release/2.3.0-notes.rst b/doc/source/release/2.3.0-notes.rst index 74f11a0b4537..d6c118a3e477 100644 --- a/doc/source/release/2.3.0-notes.rst +++ b/doc/source/release/2.3.0-notes.rst @@ -4,16 +4,519 @@ NumPy 2.3.0 Release Notes ========================== +The NumPy 2.3.0 release continues the work to improve free threaded Python +support and annotations together with the usual set of bug fixes. It is unusual +in the number of expired deprecations and the number of code modernizations and +style cleanups. The latter may not be visible to users, but is important for +code maintenance over the long term. Note that we have also upgraded from +manylinux2014 to manylinux_2_28. + +There are known test failures in the rc1 release involving MyPy and PyPy. The +cause of both has been determined and fixes will be applied before the final +release. The current Windows on ARM wheels also lack OpenBLAS, but they should +suffice for initial downstream testing. OpenBLAS will be incorporated in those +wheels when it becomes available. + +This release supports Python versions 3.11-3.13, Python 3.14 will +be supported when it is released. + Highlights ========== -*We'll choose highlights for this release near the end of the release cycle.* +* Interactive examples in the NumPy documentation. +* Building NumPy with OpenMP Parallelization. +* Preliminary support for Windows on ARM. +* Improved support for free threaded Python. +* Improved annotations. + + +New functions +============= + +New function ``numpy.strings.slice`` +------------------------------------ +The new function ``numpy.strings.slice`` was added, which implements fast +native slicing of string arrays. It supports the full slicing API including +negative slice offsets and steps. + +(`gh-27789 `__) + + +Deprecations +============ + +* The ``numpy.typing.mypy_plugin`` has been deprecated in favor of platform-agnostic + static type inference. Please remove ``numpy.typing.mypy_plugin`` from the ``plugins`` + section of your mypy configuration. If this change results in new errors being + reported, kindly open an issue. + + (`gh-28129 `__) + +* The ``numpy.typing.NBitBase`` type has been deprecated and will be removed in + a future version. + + This type was previously intended to be used as a generic upper bound for + type-parameters, for example: + + .. code-block:: python + + import numpy as np + import numpy.typing as npt + + def f[NT: npt.NBitBase](x: np.complexfloating[NT]) -> np.floating[NT]: ... + + But in NumPy 2.2.0, ``float64`` and ``complex128`` were changed to concrete + subtypes, causing static type-checkers to reject ``x: np.float64 = + f(np.complex128(42j))``. + + So instead, the better approach is to use ``typing.overload``: + + .. code-block:: python + + import numpy as np + from typing import overload + + @overload + def f(x: np.complex64) -> np.float32: ... + @overload + def f(x: np.complex128) -> np.float64: ... + @overload + def f(x: np.clongdouble) -> np.longdouble: ... + + (`gh-28884 `__) + + +Expired deprecations +==================== + +* Remove deprecated macros like ``NPY_OWNDATA`` from Cython interfaces in favor + of ``NPY_ARRAY_OWNDATA`` (deprecated since 1.7) + + (`gh-28254 `__) + +* Remove ``numpy/npy_1_7_deprecated_api.h`` and C macros like ``NPY_OWNDATA`` + in favor of ``NPY_ARRAY_OWNDATA`` (deprecated since 1.7) + + (`gh-28254 `__) + +* Remove alias ``generate_divbyzero_error`` to + ``npy_set_floatstatus_divbyzero`` and ``generate_overflow_error`` to + ``npy_set_floatstatus_overflow`` (deprecated since 1.10) + + (`gh-28254 `__) + +* Remove ``np.tostring`` (deprecated since 1.19) + + (`gh-28254 `__) + +* Raise on ``np.conjugate`` of non-numeric types (deprecated since 1.13) + + (`gh-28254 `__) + +* Raise when using ``np.bincount(...minlength=None)``, use 0 instead + (deprecated since 1.14) + + (`gh-28254 `__) + +* Passing ``shape=None`` to functions with a non-optional shape argument + errors, use ``()`` instead (deprecated since 1.20) + + (`gh-28254 `__) + +* Inexact matches for ``mode`` and ``searchside`` raise (deprecated since 1.20) + + (`gh-28254 `__) + +* Setting ``__array_finalize__ = None`` errors (deprecated since 1.23) + + (`gh-28254 `__) + +* ``np.fromfile`` and ``np.fromstring`` error on bad data, previously they + would guess (deprecated since 1.18) + + (`gh-28254 `__) + +* ``datetime64`` and ``timedelta64`` construction with a tuple no longer + accepts an ``event`` value, either use a two-tuple of (unit, num) or a + 4-tuple of (unit, num, den, 1) (deprecated since 1.14) + + (`gh-28254 `__) + +* When constructing a ``dtype`` from a class with a ``dtype`` attribute, that + attribute must be a dtype-instance rather than a thing that can be parsed as + a dtype instance (deprecated in 1.19). At some point the whole construct of + using a dtype attribute will be deprecated (see #25306) + + (`gh-28254 `__) + +* Passing booleans as partition index errors (deprecated since 1.23) + + (`gh-28254 `__) + +* Out-of-bounds indexes error even on empty arrays (deprecated since 1.20) + + (`gh-28254 `__) + +* ``np.tostring`` has been removed, use ``tobytes`` instead (deprecated since 1.19) + + (`gh-28254 `__) + +* Disallow make a non-writeable array writeable for arrays with a base that do + not own their data (deprecated since 1.17) + + (`gh-28254 `__) + +* ``concatenate()`` with ``axis=None`` uses ``same-kind`` casting by default, + not ``unsafe`` (deprecated since 1.20) + + (`gh-28254 `__) + +* Unpickling a scalar with object dtype errors (deprecated since 1.20) + + (`gh-28254 `__) + +* The binary mode of ``fromstring`` now errors, use ``frombuffer`` instead + (deprecated since 1.14) + + (`gh-28254 `__) + +* Converting ``np.inexact`` or ``np.floating`` to a dtype errors (deprecated + since 1.19) + + (`gh-28254 `__) + +* Converting ``np.complex``, ``np.integer``, ``np.signedinteger``, + ``np.unsignedinteger``, ``np.generic`` to a dtype errors (deprecated since + 1.19) + + (`gh-28254 `__) + +* The Python built-in ``round`` errors for complex scalars. Use ``np.round`` or + ``scalar.round`` instead (deprecated since 1.19) + + (`gh-28254 `__) + +* 'np.bool' scalars can no longer be interpreted as an index (deprecated since 1.19) + + (`gh-28254 `__) + +* Parsing an integer via a float string is no longer supported. (deprecated + since 1.23) To avoid this error you can + * make sure the original data is stored as integers. + * use the ``converters=float`` keyword argument. + * Use ``np.loadtxt(...).astype(np.int64)`` + + (`gh-28254 `__) + +* The use of a length 1 tuple for the ufunc ``signature`` errors. Use ``dtype`` + or fill the tuple with ``None`` (deprecated since 1.19) + + (`gh-28254 `__) + +* Special handling of matrix is in np.outer is removed. Convert to a ndarray + via ``matrix.A`` (deprecated since 1.20) + + (`gh-28254 `__) + +* Removed the ``np.compat`` package source code (removed in 2.0) + + (`gh-28961 `__) + + +C API changes +============= + +* ``NpyIter_GetTransferFlags`` is now available to check if + the iterator needs the Python API or if casts may cause floating point + errors (FPE). FPEs can for example be set when casting ``float64(1e300)`` + to ``float32`` (overflow to infinity) or a NaN to an integer (invalid value). + + (`gh-27883 `__) + +* ``NpyIter`` now has no limit on the number of operands it supports. + + (`gh-28080 `__) + +New ``NpyIter_GetTransferFlags`` and ``NpyIter_IterationNeedsAPI`` change +------------------------------------------------------------------------- +NumPy now has the new ``NpyIter_GetTransferFlags`` function as a more precise +way checking of iterator/buffering needs. I.e. whether the Python API/GIL is +required or floating point errors may occur. +This function is also faster if you already know your needs without buffering. + +The ``NpyIter_IterationNeedsAPI`` function now performs all the checks that were +previously performed at setup time. While it was never necessary to call it +multiple times, doing so will now have a larger cost. + +(`gh-27998 `__) + + +New Features +============ + +* The type parameter of ``np.dtype`` now defaults to ``typing.Any``. + This way, static type-checkers will infer ``dtype: np.dtype`` as + ``dtype: np.dtype[Any]``, without reporting an error. + + (`gh-28669 `__) + +* Static type-checkers now interpret: + + - ``_: np.ndarray`` as ``_: npt.NDArray[typing.Any]``. + - ``_: np.flatiter`` as ``_: np.flatiter[np.ndarray]``. + + This is because their type parameters now have default values. + + (`gh-28940 `__) + +NumPy now registers its pkg-config paths with the pkgconf_ PyPI package +----------------------------------------------------------------------- +The pkgconf_ PyPI package provides an interface for projects like NumPy to +register their own paths to be added to the pkg-config search path. This means +that when using pkgconf_ from PyPI, NumPy will be discoverable without needing +for any custom environment configuration. + +.. attention:: Attention + + This only applies when using the pkgconf_ package from PyPI_, or put another + way, this only applies when installing pkgconf_ via a Python package + manager. + + If you are using ``pkg-config`` or ``pkgconf`` provided by your system, or + any other source that does not use the pkgconf-pypi_ project, the NumPy + pkg-config directory will not be automatically added to the search path. In + these situations, you might want to use ``numpy-config``. + + +.. _pkgconf: https://github.com/pypackaging-native/pkgconf-pypi +.. _PyPI: https://pypi.org/ +.. _pkgconf-pypi: https://github.com/pypackaging-native/pkgconf-pypi + +(`gh-28214 `__) + +Allow ``out=...`` in ufuncs to ensure array result +-------------------------------------------------- +NumPy has the sometimes difficult behavior that it currently usually +returns scalars rather than 0-D arrays (even if the inputs were 0-D arrays). +This is especially problematic for non-numerical dtypes (e.g. ``object``). + +For ufuncs (i.e. most simple math functions) it is now possible to use +``out=...`` (literally \`...\`, e.g. ``out=Ellipsis``) which is identical in +behavior to ``out`` not being passed, but will ensure a non-scalar return. +This spelling is borrowed from ``arr1d[0, ...]`` where the ``...`` also ensures +a non-scalar return. + +Other functions with an ``out=`` kwarg should gain support eventually. +Downstream libraries that interoperate via ``__array_ufunc__`` or +``__array_function__`` may need to adapt to support this. + +(`gh-28576 `__) + +Building NumPy with OpenMP Parallelization +------------------------------------------ +NumPy now supports OpenMP parallel processing capabilities when built with the +``-Denable_openmp=true`` Meson build flag. This feature is disabled by default. +When enabled, ``np.sort`` and ``np.argsort`` functions can utilize OpenMP for +parallel thread execution, improving performance for these operations. + +(`gh-28619 `__) + +Interactive examples in the NumPy documentation +----------------------------------------------- +The NumPy documentation includes a number of examples that +can now be run interactively in your browser using WebAssembly +and Pyodide. + +Please note that the examples are currently experimental in +nature and may not work as expected for all methods in the +public API. + +(`gh-26745 `__) + + +Improvements +============ + +* Scalar comparisons between non-comparable dtypes such as + ``np.array(1) == np.array('s')`` now return a NumPy bool instead of + a Python bool. + + (`gh-27288 `__) + +* ``np.nditer`` now has no limit on the number of supported operands + (C-integer). + + (`gh-28080 `__) + +* No-copy pickling is now supported for any + array that can be transposed to a C-contiguous array. + + (`gh-28105 `__) + +* The ``__repr__`` for user-defined dtypes now prefers the ``__name__`` of the + custom dtype over a more generic name constructed from its ``kind`` and + ``itemsize``. + + (`gh-28250 `__) + +* ``np.dot`` now reports floating point exceptions. + + (`gh-28442 `__) + +* ``np.dtypes.StringDType`` is now a + `generic type `_ which + accepts a type argument for ``na_object`` that defaults to ``typing.Never``. + For example, ``StringDType(na_object=None)`` returns a ``StringDType[None]``, + and ``StringDType()`` returns a ``StringDType[typing.Never]``. + + (`gh-28856 `__) + +Added warnings to ``np.isclose`` +-------------------------------- +Added warning messages if at least one of atol or rtol are either ``np.nan`` or +``np.inf`` within ``np.isclose``. + +* Warnings follow the user's ``np.seterr`` settings + +(`gh-28205 `__) + + +Performance improvements and changes +==================================== + +Performance improvements to ``np.unique`` +----------------------------------------- +``np.unique`` now tries to use a hash table to find unique values instead of +sorting values before finding unique values. This is limited to certain dtypes +for now, and the function is now faster for those dtypes. The function now also +exposes a ``sorted`` parameter to allow returning unique values as they were +found, instead of sorting them afterwards. + +(`gh-26018 `__) + +Performance improvements to ``np.sort`` and ``np.argsort`` +---------------------------------------------------------- +``np.sort`` and ``np.argsort`` functions now can leverage OpenMP for parallel +thread execution, resulting in up to 3.5x speedups on x86 architectures with +AVX2 or AVX-512 instructions. This opt-in feature requires NumPy to be built +with the -Denable_openmp Meson flag. Users can control the number of threads +used by setting the OMP_NUM_THREADS environment variable. + +(`gh-28619 `__) + +Performance improvements for ``np.float16`` casts +------------------------------------------------- +Earlier, floating point casts to and from ``np.float16`` types +were emulated in software on all platforms. + +Now, on ARM devices that support Neon float16 intrinsics (such as +recent Apple Silicon), the native float16 path is used to achieve +the best performance. + +(`gh-28769 `__) + + +Changes +======= + +* The vector norm ``ord=inf`` and the matrix norms ``ord={1, 2, inf, 'nuc'}`` + now always returns zero for empty arrays. Empty arrays have at least one axis + of size zero. This affects ``np.linalg.norm``, ``np.linalg.vector_norm``, and + ``np.linalg.matrix_norm``. Previously, NumPy would raises errors or return + zero depending on the shape of the array. + + (`gh-28343 `__) + +* A spelling error in the error message returned when converting a string to a + float with the method ``np.format_float_positional`` has been fixed. + + (`gh-28569 `__) + +* NumPy's ``__array_api_version__`` was upgraded from ``2023.12`` to ``2024.12``. + +* ``numpy.count_nonzero`` for ``axis=None`` (default) now returns a NumPy scalar + instead of a Python integer. + +* The parameter ``axis`` in ``numpy.take_along_axis`` function has now a default + value of ``-1``. + + (`gh-28615 `__) + +* Printing of ``np.float16`` and ``np.float32`` scalars and arrays have been improved by + adjusting the transition to scientific notation based on the floating point precision. + A new legacy ``np.printoptions`` mode ``'2.2'`` has been added for backwards compatibility. + + (`gh-28703 `__) + +``unique_values`` may return unsorted data +------------------------------------------ +The relatively new function (added in NumPy 2.0) ``unique_values`` may now +return unsorted results. Just as ``unique_counts`` and ``unique_all`` these +never guaranteed a sorted result, however, the result was sorted until now. In +cases where these do return a sorted result, this may change in future releases +to improve performance. + +(`gh-26018 `__) + +Changes to the main iterator and potential numerical changes +------------------------------------------------------------ +The main iterator, used in math functions and via ``np.nditer`` from Python and +``NpyIter`` in C, now behaves differently for some buffered iterations. This +means that: + +* The buffer size used will often be smaller than the maximum buffer sized + allowed by the ``buffersize`` parameter. + +* The "growinner" flag is now honored with buffered reductions when no operand + requires buffering. + +For ``np.sum()`` such changes in buffersize may slightly change numerical +results of floating point operations. Users who use "growinner" for custom +reductions could notice changes in precision (for example, in NumPy we removed +it from ``einsum`` to avoid most precision changes and improve precision for +some 64bit floating point inputs). + +(`gh-27883 `__) + +The minimum supported GCC version is now 9.3.0 +---------------------------------------------- +The minimum supported version was updated from 8.4.0 to 9.3.0, primarily in +order to reduce the chance of platform-specific bugs in old GCC versions from +causing issues. + +(`gh-28102 `__) + +Changes to automatic bin selection in numpy.histogram +----------------------------------------------------- +The automatic bin selection algorithm in ``numpy.histogram`` has been modified +to avoid out-of-memory errors for samples with low variation. For full control +over the selected bins the user can use set the ``bin`` or ``range`` parameters +of ``numpy.histogram``. + +(`gh-28426 `__) + +Build manylinux_2_28 wheels +--------------------------- +Wheels for linux systems will use the ``manylinux_2_28`` tag (instead of the +``manylinux2014`` tag), which means dropping support for redhat7/centos7, +amazonlinux2, debian9, ubuntu18.04, and other pre-glibc2.28 operating system +versions, as per the `PEP 600 support table`_. + +.. _`PEP 600 support table`: https://github.com/mayeut/pep600_compliance?tab=readme-ov-file#pep600-compliance-check + +(`gh-28436 `__) +Remove use of -Wl,-ld_classic on macOS +-------------------------------------- +Remove use of -Wl,-ld_classic on macOS. This hack is no longer needed by Spack, +and results in libraries that cannot link to other libraries built with ld +(new). -.. if release snippets have been incorporated already, uncomment the follow - line (leave the `.. include:: directive) +(`gh-28713 `__) -.. **Content from release note snippets in doc/release/upcoming_changes:** +Re-enable overriding functions in the ``numpy.strings`` +------------------------------------------------------- +Re-enable overriding functions in the ``numpy.strings`` module. -.. include:: notes-towncrier.rst +(`gh-28741 `__) diff --git a/pyproject.toml b/pyproject.toml index b62d71cbba73..17fbf71088b8 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -7,7 +7,7 @@ requires = [ [project] name = "numpy" -version = "2.3.0.dev0" +version = "2.3.0rc1" # TODO: add `license-files` once PEP 639 is accepted (see meson-python#88) license = {file = "LICENSE.txt"}