v2026.07.0

This release adds support for Dask's query-optimizing expression arrays, along
with new day_of_week and day_of_year datetime accessor attributes. It
also includes a number of bug fixes, notably for a performance regression in
:py:meth:Coordinates.to_index, Zarr fill_value round-tripping, and
excessive memory use in drop_encoding.

Thanks to the 25 contributors to this release:
Davis Bennett, Deepak Cherian, Ian Hunt-Isaak, Illviljan, Jonathan Dung, Julia
Signell, Justus Magin, Kai Mühlbauer, MJSHANG, Mark Harfouche, Mathias Hauser,
Matt Van Horn, Matthew Rocklin, Max Jones, Maximilian Roos, Nick Hodgskin,
S Anand, Spencer Clark, Sreekant Baheti, Timothy Hodson, Tom Nicholas,
Vincent Gao, Wali Reheman, Wei Ji and eeshsaxena

What's Changed

• Empty section at top of whats-new by @jsignell in #11299
• Fix performance regression in Coordinates.to_index by @thodson-usgs in #11306
• Document Dataset.load_async and DataArray.load_async by @TomNicholas in #11311
• Fix assert_equal with check_dim_order=False for mixed dimension orders by @max-sixty in #10718
• work around a bug in pixi that disallows custom PyPI servers by @keewis in #11331
• undo the pin on the pixi version by @keewis in #11334
• Objects satisfying the core.types.Lock interface should also be hashable by @jonathandung in #11333
• fix: ensure fill_value for zarr v3 arrays round-trips by @d-v-b in #11345
• Mark test_roundtrip_pandas_dataframe_datetime as an expected failure by @spencerkclark in #11365
• Fix mypy failure: drop dead zarr<3 branch in invalid-store test by @hmaarrfk in #11366
• ci: move pandas/scipy stubs to pypi deps by @maxrjones in #11370
• Fix Upstream CI NaT issues by @ianhi in #11340
• pin pixi until the nightly ci resolves again by @keewis in #11372
• refactor: remove dead zarr<3 (has_zarr_v3) guards in tests by @mvanhorn in #11376
• Preserve step in RangeIndex.arange and slicing by @mokashang in #11362
• DOC: Clarify docstring of Dataset.load by @VeckoTheGecko in #11388
• Use to_numpy in ImplicitToExplicitIndexingAdapter.array method by @weiji14 in #11381
• ci: add Python 3.14 test environment to CI matrix by @mvanhorn in #11377
• Silence datetime accessor deprecation warnings by @spencerkclark in #11270
• Support query optimization with Dask expression arrays by @mrocklin in #11382
• debug the nightly CI by @keewis in #11395
• Add nodask constraint for test-py314-no-dask environment by @weiji14 in #11385
• Fix RangeIndex.linspace with num=1 by @sanand0 in #11401
• Fix drop_encoding to avoid copying data (GH#11390) by @wali-reheman in #11394
• fix: handle OSError in _get_mtime for non-file paths by @gaoflow in #11392
• Add dask expr protocol for variables by @mrocklin in #11423
• Add map_blocks and some testing support for array query expressions by @mrocklin in #11398
• Fix incorrect parameter name in cumulative() docstrings by @Sreekant13 in #11425
• MAINT: remove pandas 2.2 compat (incl. infer_freq) by @mathause in #11430
• Adapt to coming pd.infer_freq behavior change by @spencerkclark in #11434
• Adapt test_dataset_math_errors for coming NumPy behavior change by @spencerkclark in #11435
• DOC fix parameter name in CoordinateTransform.reverse docstring by @eeshsaxena in #11433
• open an issue even with a missing log file by @keewis in #11436
• zarr deprecation: enums becoming string literals by @keewis in #11438
• resolve the sparse errors by @keewis in #11439
• Release notes for v2026.07.0 by @shoyer in #11445

New Contributors

• @thodson-usgs made their first contribution in #11306
• @jonathandung made their first contribution in #11333
• @mvanhorn made their first contribution in #11376
• @mokashang made their first contribution in #11362
• @sanand0 made their first contribution in #11401
• @wali-reheman made their first contribution in #11394
• @gaoflow made their first contribution in #11392
• @Sreekant13 made their first contribution in #11425
• @eeshsaxena made their first contribution in #11433

Full Changelog: v2026.04.0...v2026.07.0