Skip to content

Releases: scikit-hep/awkward

Version 2.6.8

12 Sep 21:01
dea122a
Compare
Choose a tag to compare

New features

Bug-fixes and performance

  • fix: 2881 akconcatenate fails trying to concatenate too many nested arrays by @tcawlfield in #3207
  • fix: adjust checks in ForthMachine to prevent segfault when num_items is negative by @ariostas in #3209
  • fix: generalize Index.ptr by @jpivarski in #3206
  • fix: add ak.array_equal to NEP18 overrides, documentation, and add one more test by @tcawlfield in #3225
  • fix: cuda max reducer bugfix and more tests by @ianna in #3228
  • fix: empty arrays in ak.to_parquet with extensionarray=True by @jpivarski in #3234
  • fix: GPU complex reducer prod for empty lists by @ianna in #3235

Other

New Contributors

Full Changelog: v2.6.7...v2.6.8

Version 2.6.7

02 Aug 19:23
3ab54f8
Compare
Choose a tag to compare

New features

Bug-fixes and performance

Other

New Contributors

Full Changelog: v2.6.6...v2.6.7

Version 2.6.6

26 Jun 18:57
7fce3d4
Compare
Choose a tag to compare

New features

Bug-fixes and performance

  • fix: use grid-stride looping for kernels with variable-length loops by @ManasviGoyal in #3130
  • fix: don't let CuPy iterate over Index with Python for loops by @jpivarski in #3142
  • fix: correct dtypes for numpy v2 by @ianna in #3159
  • fix: ak.from_numpy should fail on zero-dimensional arrays. by @tcawlfield in #3161

Other

Full Changelog: v2.6.5...v2.6.6

Version 2.6.5

28 May 20:18
45c708d
Compare
Choose a tag to compare

New features

  • feat: 2772 pyarrow doesnt permit selective reading with extensionarray by @tcawlfield in #3127

Bug-fixes and performance

  • fix: skip failing test when jax is not installed by @ianna in #3112
  • fix: skip tests that require pyarrow when it is not installed by @ianna in #3114
  • fix: generate array view by @ianna in #3115
  • fix: Header-only fix is_valid for empty Indexed or IndexedOption by @zonca in #3120
  • fix: wait for Jitify performing a one-time only warm-up by @ianna in #3113
  • fix: prevent exponential memory growth in UnionArray by @jpivarski in #3119

Other

Full Changelog: v2.6.4...v2.6.5

Version 2.6.4

03 May 22:27
836b28a
Compare
Choose a tag to compare

New features

Bug-fixes and performance

  • fix: restrict ak.softmax to axis=-1, the only axis for which the calculation is correct right now by @jpivarski in #3072
  • fix: Numpy 2.0 out of bounds error for BitMaskedArray Py kernels by @ManasviGoyal in #3073
  • fix: _is_getitem_at_placeholder logic was backward for UnionArray by @jpivarski in #3083
  • fix: refuse dak.Array in hidden-but-public-functions by @jpivarski in #3079
  • fix: PyPy segfault may be related to using out-of-scope array by @jpivarski in #3093
  • fix: distributed rdataframe is not yet support by @ianna in #3087
  • fix: support conversion to arrow and back with non-option Unknown type by @tcawlfield in #3085
  • fix: intermittent segfault in PyPy by @ianna in #3089
  • fix: is_valid checks in header-only library by @zonca in #3091
  • fix: UnmaskedArray was missing its _nextcarry_outindex by @jpivarski in #3097

Other

Full Changelog: v2.6.3...v2.6.4

Version 2.6.3

01 Apr 23:01
8c39944
Compare
Choose a tag to compare

This is the first release of Awkward Array that is compatible with NumPy 2.0.

New features

  • feat!: printing a typetracer array should not touch it by @jpivarski in #3019
  • feat: add to_parquet_dataset function by @zbilodea in #2898
  • feat: adding support for numpy.real, imag, round, angle, by @tcawlfield in #3053
  • feat: add miscellaneous CUDA kernels by @ManasviGoyal in #3021
  • feat: add dask_tokenize hook by @lgray in #3017
  • feat: added a commit method on TypeTracerReport to identify touched buffers in the Dask DAG-building pass by @jpivarski in #3043
  • feat: Allow interoperability with numpy 2.x by @tcawlfield in #3064

Bug-fixes and performance

  • fix: return ufunc as it is for user defined vectorized funcs by @Saransh-cpp in #3025
  • fix: Support structured arrays in option type to_numpy by @zonca in #3060
  • fix: Windows and 32-bit Linux default to 32-bit integers in sums and prods UNLESS it's NumPy 2.0 by @jpivarski in #3068
  • perf: add a flag to turn off isinstance in TypeTracerArray._new by @jpivarski in #3054

Other

New Contributors

Full Changelog: v2.6.2...v2.6.3

Version 2.6.3rc2

22 Mar 13:29
958deb9
Compare
Choose a tag to compare
Version 2.6.3rc2 Pre-release
Pre-release

New features

  • feat!: printing a typetracer array should not touch it by @jpivarski in #3019
  • feat: add to_parquet_dataset function by @zbilodea in #2898
  • feat: adding support for numpy.real, imag, round, angle, by @tcawlfield in #3053
  • feat: add miscellaneous CUDA kernels by @ManasviGoyal in #3021
  • feat: add dask_tokenize hook by @lgray in #3017
  • feat: added a commit method on TypeTracerReport to identify touched buffers in the Dask DAG-building pass by @jpivarski in #3043

Bug-fixes and performance

  • fix: return ufunc as it is for user defined vectorized funcs by @Saransh-cpp in #3025
  • perf: add a flag to turn off isinstance in TypeTracerArray._new by @jpivarski in #3054

Other

New Contributors

Full Changelog: v2.6.2...v2.6.3rc2

Version 2.6.3rc1

21 Mar 22:31
67b641b
Compare
Choose a tag to compare
Version 2.6.3rc1 Pre-release
Pre-release

The main purpose of this release is to make sure that the deployment machinery still works, now that the PYPI_PASSWORD had to be replaced.

New features

Bug-fixes and performance

  • fix: return ufunc as it is for user defined vectorized funcs by @Saransh-cpp in #3025

Other

New Contributors

Full Changelog: v2.6.2...v2.6.3rc1

Version 2.6.2

04 Mar 23:59
1656cd7
Compare
Choose a tag to compare

New features

Bug-fixes and performance

  • fix: remove redundant(?) Jax.is_tracer_type check in _layout by @Saransh-cpp in #3013
  • fix: throw more meaningful error message for non-NumPy buffers by @agoose77 in #3026
  • fix: proper typetracer array slicing in BitMaskedArray by @agoose77 in #3028
  • fix: ak.flatten on BitMaskedArray by @agoose77 in #3034
  • fix: performance of ufunc resolution for non-nominal signatures by @agoose77 in #3030

Other

Full Changelog: v2.6.1...v2.6.2

Version 2.6.1

05 Feb 21:35
Compare
Choose a tag to compare

The purpose of this release is to provide a lower-bound on the fsspec dependency version (#3006).

New features

(none!)

Bug-fixes and performance

(none!)

Other

Full Changelog: v2.6.0...v2.6.1