Version 2.7.0

Features:

• Add a function to integrate axes #505
• Add fill_flattened support #474

Various other items:

• Move linting to using Ruff #475
• Self & _compat #479
• Rework and rerun docs API #481

New Contributors

• @fabriceMUKARAGE made their first contribution in #502
• @agoose77 made their first contribution in #474

Full Changelog: v2.6.3...v2.7.0

Version 2.6.3

• Experimental dask support #471
• Some minor cleanups from refurb #453

New Contributors

• @lgray made their first contribution in #471

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

Version 2.6.2

• Nicer stacks repr #449
• Backport storage_type if boost-histogram < 1.3.2 #447
• Allow overwriting labels for plot/overlay #414
• Use Hatching to build the package #418
• Support git archival version numbers #441

Version 2.6.1

• Fall back on normal repr when histogram is too large #388
• Fix issue with no-axis histogram #388
• Fix issue with empty axis causing segfault until fixed upstream #387
• Only require SciPy if using SciPy #386

Version 2.6.0

• Using boost-histogram 1.3
• Fix runtime dependency on matplotlib when not plotting #353
• Fix .plot shortcut failure #368
• New nox sessions: regenerate and pylint
• Update tests for latest matplotlib

Version 2.5.2

• Remove more-itertools requirement #347
• Fix missing pass-through for stack plot #339

Version 2.5.1

• Support named stack indexing #325
• Fix histoprint error with stacks #325
• Better README

Version 2.5.0

• Stacks support axes, math operations, projection, setting items, and iter/dict construction. They also support histogram titles in legends. Added histoprint support for Stacks. #291 #315 #317 #318
• Added name= and label= to histograms, include Hist arguments in QuickConstruct. #297
• AxesTuple now supports bulk name setting, h.axes.name = ("a", "b", ...). #288
• Added hist.new alias for hist.Hist.new. #296
• Added "efficiency" uncertainty_type option for ratio_plot API. #266 #278

Smaller features or fixes:

• Dropped Python 3.6 support. #194
• Uses boost-histogram 1.2.x series, includes all features and fixes, and Python 3.10 support.
• No longer require scipy or iminuit unless actually needed. #316
• Improve and clarify treatment of confidence intervals in intervals submodule. #281
• Use NumPy 1.21 for static typing. #285
• Support running tests without plotting requirements. #321

Version 2.4.0

• Support .stack(axis) and stacked histograms. #244 #257 #258
• Support selection lists (experimental with boost-histogram 1.1.0). #255
• Support full names for QuickConstruct, and support mistaken usage in constructor. #256
• Add .sort(axis) for quickly sorting a categorical axis. #243

Smaller features or fixes:

• Support nox for easier contributor setup. #228
• Better name axis error. #232
• Fix for issue plotting size 0 axes. #238
• Fix issues with repr information missing. #241
• Fix issues with wrong plot shortcut being triggered by Integer axes. #247
• Warn and better error if overlapping keyword used as axis name. #250

Along with lots of smaller docs updates.

Version 2.3.0

• Add plot_ratio to the public API, which allows for making ratio plots between the histogram and either another histogram or a callable. #161
• Add .profile to compute a (N-1)D profile histogram. #160
• Support plot1d / plot on Histograms with a categorical axis. #174
• Add frequentist coverage interval support in the intervals module. #176
• Allow plot_pull to take a more generic callable or a string as a fitting function. Introduce an option to perform a likelihood fit. Write fit parameters' values and uncertainties in the legend. #149
• Add fit_fmt= to plot_pull to control display of fit params. #168
• Support <prefix>_kw arguments for setting each axis in plotting. #193
• Cleaner IPython completion for Python 3.7+. #179