Skip to content

Version 2.8.2

Latest
Compare
Choose a tag to compare
@ianna ianna released this 03 May 10:36
6f8fb8c

New features

  • feat: add segmented argmin and argmax by @ianna in #3449
  • feat: change argmin and argmax algorithms to handle big data by @ianna in #3450
  • feat: implement segment count_nonzero reducer by @ianna in #3452
  • feat: support unknown_length for virtual arrays in order to read without any materialization by @ikrommyd in #3475
  • feat: enable jax backend for virtual arrays by @ikrommyd in #3451
  • feat: Add a parquet uuid calculation by @NJManganelli in #3440

Bug-fixes and performance

  • fix: typo in UnionArray combinations by @ikrommyd in #3438
  • fix: pyarrow 19.0 support by @ianna in #3416
  • fix: force gc in numba test by @ianna in #3448
  • fix: highlevel function dispatch type hint by @pfackeldey in #3447
  • fix: remove mixed numpy<->jax arrays in Jax backend by @pfackeldey in #3446
  • fix: better support for pyright by @henryiii in #3453
  • fix: setitem in ak firsts with jax backend by @pfackeldey in #3455
  • fix: replace some __setitem__ calls with .at[where].set(value) for the jax backend by @ikrommyd in #3457
  • fix: remove duplicate failing raggedtensor test when you have a GPU on the system by @ikrommyd in #3461
  • fix: single int slicing for named axis by @pfackeldey in #3469
  • fix: add ptr implementation and fix __setitem__ for Index and tocarryraw to make ak.(arg)combinations work for the jax backend by @ikrommyd in #3471
  • fix: check for error being None in the output of validity kernels by @ikrommyd in #3479
  • fix: minor things required for the whole ci to work with virtual arrays by @ikrommyd in #3484
  • fix: jax reducers returning incorrect output values or lengths by @ikrommyd in #3464
  • fix: avoid calling shape_generator inside generator if it has already been called by @ikrommyd in #3490
  • fix: add field metadata to record its option type by @ianna in #3478
  • fix: add maps_as_pydicts parameter to to_pylist by @ianna in #3492
  • fix: consistent materializing length across all layouts by @ikrommyd in #3489

Other

New Contributors

Full Changelog: v2.8.1...v2.8.2