2023.5.0

What's Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #79
• fix/feat: Mockable reduction layer by @lgray in #80
• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #81

Full Changelog: 2023.4.4...2023.5.0

2023.4.4

What's Changed

• [pre-commit.ci] pre-commit autoupdate by @pre-commit-ci in #77
• fix: axes name assignments should persist through .copy() and .compute() by @lgray in #78

Full Changelog: 2023.4.3...2023.4.4

2023.4.3

What's Changed

• feat: use upstream DataFrameTreeReduction for reducing histograms by @douglasdavis in #74
• allow release to trigger publish by @lgray in #75

Full Changelog: 2023.4.2...2023.4.3

2023.4.2

Release using GitHub actions

(no changes to the Python package)

2023.4.1

What's Changed

• Make metadata for boost histograms more lightweight by @lgray in #63

Full Changelog: 2023.4.0...2023.4.1

2023.4.0

New:

• feat: support float weight argument by @douglasdavis in #69
• docs: add release docs by @douglasdavis in #70

Misc

• pre-commit autoupdates

Full Changelog: 2023.2.0...2023.4.0

2023.2.0

New

• Support for weights= and sample= when filling with dask-awkward arrays (#57)
• Make split_every configurable (aggregation of partitioned histograms to a single histogram) (#58)

Fixes

• Make sure empty histograms can be tokenized (85dd2a6)

Other

• Some pre-commit/CI updates

2023.1.4

• Remove hard dependency on pandas (by removing use of function from dask.dataframe) (#50, #51)

2023.1.3

• Make fill calls with dask-awkward input compatible with awkward typetracers (this allows future dask-awkward optimizations to work with dask-histogram layers) (#48)
• Run dask-awkward optimizations if the awkward key exists in dask.config (#49)
• Some pre-commit updates

2023.1.2

• Fix circular import (#46)