CHANGELOG.md

Changelog

This is the changelog for arro3. pyo3-arrow has a separate changelog.

[0.8.0] - 2026-02-22

Breaking Changes

• A bare DataType is now exported through __arrow_c_schema__ (such as when passing to pyarrow.field) with nullable: true. This is required to match equality semantics with pyarrow. by @kylebarron in #483

Bug Fixes

• fix!: Fix type equality between arro3 and pyarrow by @kylebarron in #483

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.7.0...py-v0.8.0

[0.7.0] - 2026-02-19

Breaking changes

• feat!: Make Table.remove_column return IndexError by @surister in #442
• fix: Rename names to columns parameter in Tables.drop_columns by @surister in #453
• fix: Properly return IndexError on incorrect index in Table.add_column by @surister in #455

Improvements

• ci: Build 3.14t wheels by @kylebarron in #428
• feat: Make Table.append_column support Array by @surister in #433
• feat: Make Table.set_column and Table.add_column support Array and ArrayReader by @surister in #437
• fix: off by one when checking if an index is valid by @surister in #443
• feat: Implement Table.drop_columns by @surister in #440
• fix: Table.rename not modifying internals batches schema by @surister in #457
• fix: Improve error messaging by adding more information by @surister in #447
• feat: Match RecordBatch.from_arrays closely to pyarrow by @AlenkaF in #461

Bug Fixes

• fix: Preserve schema metadata in RecordBatch PyCapsule export by @bretttully in #474
• fix: Update license metadata in pyproject.toml by @kylebarron in #480

Other improvements

• chore: Bump to arrow 57 by @kylebarron in #424
• chore: Bump arro3 to pyo3 0.27 by @kylebarron in #425

New Contributors

• @surister made their first contribution in #433
• @AlenkaF made their first contribution in #461
• @bretttully made their first contribution in #474

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.6.5...py-v0.7.0

[0.6.5] - 2025-10-13

• fix(ci): Ensure we use Python 3.11 interpreter for building abi3 wheels #416

[0.6.4] - 2025-10-13

What's Changed

• feat(arro3): Freethreading - module markers + CI by @H-Plus-Time in #275
• feat: Add support for Decimal scalar type conversion to Python by @kylebarron in #412
• ci: Build abi3 wheels where possible, falling back to version-specific wheels by @kylebarron in #408

New Contributors

• @H-Plus-Time made their first contribution in #275

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.6.3...py-v0.6.4

[0.6.3] - 2025-09-17

• feature(arro3-core): Maintain timezone info when converting scalar tz to datetime.datetime (#400)

[0.6.2] - 2025-09-09

• Enable sdist artifacts #398
  • sdist is now pushed to PyPI for all packages

[0.6.1] - 2025-08-22

arro3-core

Bug fixes 🐛

• fix(arro3-io): use patched parquet crate to fix building arro3-io for s390x architectures. #388. This also updates arrow crates to 56.1.0.

[0.6.0] - 2025-08-21

arro3-core

Breaking Changes 🔧

No breaking changes.

New features ✨

• Expanded dtype support in Array.from_numpy. It now additionally supports datetime64, timedelta64, fixed-width strings and bytes, variable-width strings, and object dtypes that contain str or bytes.
• Support FixedSizeBinary in Array constructor by @kylebarron in #358
• Add mask parameter to list/struct constructors by @kylebarron in #382

Bug fixes 🐛

• Preserve exceptions when Arrow PyCapsule dunder method fails by @kylebarron in #374

Updated dependencies ⬆️

• Use Rust arrow 56 by @kylebarron in #368. This includes all changes in the upstream Changelog.

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.5.1...py-v0.6.0

[0.5.1] - 2025-05-31

What's Changed

• chore: Use released pyo3-file by @kylebarron in #341
• feat: Support serializing binary view and string view array types to numpy by @kylebarron in #348
• fix(arro3-core): Add typing-extensions as a dependency for arro3-core by @kylebarron in #350
• feat(arro3-core): Improved repr for Array and ChunkedArray by @kylebarron in #351
• chore: bump dev versions for geoarrow-types and pyarrow by @kylebarron in #352

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.5.0...py-v0.5.1

[0.5.0] - 2025-05-27

New features ✨

• feat: access value_fields and inner fields by @ion-elgreco in #334
• feat(arro3-core): Basic data rendering in __repr__ by @kylebarron in #335

Bug fixes 🐛

• feat(arro3-core): Make DataType hashable by @kylebarron in #336
• test(arro3-core): Check that Schema is iterable by @kylebarron in #338

Documentation updates 📝

• Add sphinx interlinking for ArrowStreamExportable by @kylebarron in #339

New Contributors

• @ion-elgreco made their first contribution in #334

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.4.6...py-v0.5.0

[0.4.6] - 2025-03-10

Bug fixes 🐛

• Use 2_24 for aarch64 wheels by @kylebarron in #279
• Allow None as input into Array constructor by @kylebarron in #294
• Fix rendering __init__ in docs by @kylebarron in #295
• Subclass Buffer type from collections.abc.Buffer by @kylebarron in #297

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.4.5...py-v0.4.6

[0.4.5] - 2024-12-16

Bug fixes 🐛

• Build wheels for linux aarch64. #277

[0.4.4] - 2024-12-09

Bug fixes 🐛

• Raise IndexError and KeyError for invalid column access #272

[0.4.3] - 2024-11-21

What's Changed

• Build wheels for Python 3.13 by @kylebarron in #260
• Raise RuntimeWarning when compiled in debug mode by @kylebarron in #255
• Add Buffer.to_bytes by @kylebarron in #251
• Make buffers have a length by @martindurant in #252

New Contributors

• @martindurant made their first contribution in #252

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.4.2...py-0.4.3

[0.4.2] - 2024-10-14

What's Changed

• Ensure total numpy v2 support by @kylebarron in #226
• Fix segfault when releasing buffer protocol object by @kylebarron in #231

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.4.1...py-v0.4.2

[0.4.1] - 2024-10-07

What's Changed

• Support for string view and binary view data types (upgrade to arrow 53.1) by @kylebarron in #219
• Fix min/max with datetimes with timezones by @kylebarron in #220

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.4.0...py-v0.4.1

[0.4.0] - 2024-10-03

Enhancements 🪄

• Zero-copy buffer protocol data import by @kylebarron in #204

  • Handle multi-dimensional buffer protocol input by @kylebarron in #208
  • Prefer zero-copy in from_numpy by @kylebarron in #214

• New compute functions:

  • Add date_part by @kylebarron in #202
  • Aggregate functions: array min, max, sum by @kylebarron in #193
  • Arith functions by @kylebarron in #194
  • More compute functions by @kylebarron in #104

• Implement scalar equality by @kylebarron in #205

• Improved docstrings by @kylebarron in #198

• Add installation section to readme by @kylebarron in #189

Bug fixes 🐛

• Fix data type bit width by @kylebarron in #190
• Fix return type of RecordBatch.column by @kylebarron in #191

New Contributors

• @3ok made their first contribution in #210

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.3.1...py-v0.4.0

[0.3.1] - 2024-09-11

Bug fixes 🐛

• Enable IPC compression by @kylebarron in #187

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.3.0...py-v0.3.1

[0.3.0] - 2024-08-27

Enhancements 🪄

• Wheels for pyodide are auto-built on CI and published to Github releases.
• Ensure Parquet schema metadata is added to arrow table by @kylebarron in #137
• Access dictionary array keys and values by @kylebarron in #139
• Support reading Parquet from file objects by @kylebarron in #142
• Implement dictionary encoding by @kylebarron in #136
• Support for Arrow scalars and converting to Python objects by @kylebarron in #159
• Move functional accessors to core module by @kylebarron in #151
• Support for python buffer protocol by @kylebarron in #156

Bug fixes 🐛

• Fix field metadata propagation by @kylebarron in #150
• Set strip=true for maturin builds by @kylebarron in #155
• Support __getitem__ with a negative index by @kylebarron in #171
• support f16 in from_numpy by @kylebarron in #154
• Fix writing to file by @kylebarron in #138

Full Changelog: https://github.com/kylebarron/arro3/compare/py-v0.2.1...py-v0.3.0

[0.2.1] - 2024-08-13

Bug fixes 🐛

• Fix API docs (#131)

[0.2.0] - 2024-08-13

Enhancements 🪄

• In general, high parity with pyarrow for most data access and management. Most of the Table, ChunkedArray, Array, RecordBatchReader, Schema, and Field methods and behavior should be similar to pyarrow.
• An ArrayReader, an abstraction beyond RecordBatchReader to allow a stream of arrow arrays that are not RecordBatches.
• Readers and writers for Parquet, Arrow IPC, CSV, and JSON.
• Initial compute functions.
• Initial Python tests.
• Improved documentation, both in the type stubs and in the docs website.
• Pyodide wheel support.

[0.1.0] - 2024-07-01

• Initial release of arro3-core v0.1.