Awkward Array Files

Highlights

• Much faster list reductions (ak.sum, ak.max, …) — about 5× faster on average (geometric mean), 10×+ on long lists https://github.com/scikit-hep/awkward/pull/4056
  • Kernels rewritten: CPU (awkward-cpp), GPU (cuda.compute)
  • Exact floating-point results may differ from the previous release
  • The same arrays may have different internal layouts than before
  • Removed the undocumented ak.index.EmptyIndex / ZeroIndex / LazyIndex
• Many bug fixes and improvements from code review by Claude Fable 5 (PRs 4089–4108) https://github.com/scikit-hep/awkward/issues/4085

What's Changed

New features

• feat: complete CPU kernel migration from parents to offsets by @ianna in https://github.com/scikit-hep/awkward/pull/4056
• feat: migrate UnionArray kernels to cuda.compute by @maxymnaumchyk in https://github.com/scikit-hep/awkward/pull/4083
• feat: combine cuda.compute PRs into one by @maxymnaumchyk in https://github.com/scikit-hep/awkward/pull/4124
• feat: migrate kernels that were using parents to cuda.compute by @maxymnaumchyk in https://github.com/scikit-hep/awkward/pull/4159

Bug-fixes and performance

• fix: add missing .attrs support in parquet serialisation by @albert-physics in https://github.com/scikit-hep/awkward/pull/4079
• fix: test_pandas_attr_serialisation should have unique parquet path per thread by @ikrommyd in https://github.com/scikit-hep/awkward/pull/4110
• fix: passing a non-Collection iterable to pytest.parametrize is deprecated by @maxymnaumchyk in https://github.com/scikit-hep/awkward/pull/4144
• fix: datetime.time conversion and binary buffer copy in bindings by @henryiii in https://github.com/scikit-hep/awkward/pull/4089
• fix: properly error on conversion of IndexedOptionArray with an EmptyArray content and length > 0 to a ByteMaskedArray by @ikrommyd in https://github.com/scikit-hep/awkward/pull/4127
• fix: named-axis context sharing and ak.where broadcasting crash by @henryiii in https://github.com/scikit-hep/awkward/pull/4108
• fix: require pyarrow>=17.0.0 by @henryiii in https://github.com/scikit-hep/awkward/pull/4152
• fix: wrong results from string sort/unique kernels by @henryiii in https://github.com/scikit-hep/awkward/pull/4091
• fix: int64 precision loss in sort comparators and per-comparison branching by @henryiii in https://github.com/scikit-hep/awkward/pull/4090
• fix: LayoutBuilder Union form JSON and Record map-type support by @henryiii in https://github.com/scikit-hep/awkward/pull/4094
• fix: misc integration bugs (hist, avro, numexpr, RDataFrame) by @henryiii in https://github.com/scikit-hep/awkward/pull/4096
• fix: modify cache key for macOS deployment in workflow by @ianna in https://github.com/scikit-hep/awkward/pull/4158
• fix: reject non-integer float indices instead of truncating by @henryiii in https://github.com/scikit-hep/awkward/pull/4097
• fix: typetracer correctness issues affecting dask-awkward by @henryiii in https://github.com/scikit-hep/awkward/pull/4100
• fix: BitMaskedArray mask inversion and placeholder detection by @henryiii in https://github.com/scikit-hep/awkward/pull/4092
• fix: Arrow conversion without pandas, empty chunked arrays, and multi-file Parquet attrs by @henryiii in https://github.com/scikit-hep/awkward/pull/4099
• fix: BitMasked builder use-after-free and GrowableBuffer copy bugs by @henryiii in https://github.com/scikit-hep/awkward/pull/4098

Other

• ci: fix deploy job by @ariostas in https://github.com/scikit-hep/awkward/pull/4082
• chore: add an AGENTS file by @henryiii in https://github.com/scikit-hep/awkward/pull/4086
• chore: don't refuse to install awkward_cpp when using uv by @henryiii in https://github.com/scikit-hep/awkward/pull/4087
• chore(deps): bump the actions group with 3 updates by @dependabot[bot] in https://github.com/scikit-hep/awkward/pull/4121
• docs: add albert-physics as a contributor for code by @allcontributors[bot] in https://github.com/scikit-hep/awkward/pull/4122
• ci: automate release-note categorization with type labels by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4125
• docs: document kernel spec invariants and CPU/CUDA equivalence in AGENTS by @henryiii in https://github.com/scikit-hep/awkward/pull/4088
• ci: run property-based tests nightly with a large max_examples by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4128
• test: add a property-based kernel test with CPU and GPU backends by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4142
• ci: show Hypothesis statistics in the nightly property tests by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4146
• chore: modernize header-only code to C++17 and remove dead code by @henryiii in https://github.com/scikit-hep/awkward/pull/4104
• ci: run the GPU kernel property tests nightly at the nightly profile by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4147
• ci: enable pyodide build by @ryanking13 in https://github.com/scikit-hep/awkward/pull/4149
• ci: fix docs-preview PR comment failing without a checkout by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4155
• chore(deps): bump the actions group across 1 directory with 3 updates by @dependabot[bot] in https://github.com/scikit-hep/awkward/pull/4153
• docs: describe the returned object in Returns: for already-merged operations by @TaiSakuma in https://github.com/scikit-hep/awkward/pull/4148
• chore(deps): update python version for mypy by @maxymnaumchyk in https://github.com/scikit-hep/awkward/pull/4172

New Contributors

• @albert-physics made their first contribution in https://github.com/scikit-hep/awkward/pull/4079
• @ryanking13 made their first contribution in https://github.com/scikit-hep/awkward/pull/4149

Full Changelog: https://github.com/scikit-hep/awkward/compare/v2.9.1...v2.10.0