Skip to content

Upgrading dask 2025 to 2026

Breaking2025.12.0 → 2026.8.0Breaking3Security2Platform1Deprecated1New3Changed14Fixed23Other112

Full comparisonBack to dask
BreakingUpgrade verdict

3 breaking changes and a platform requirement change and 2 security fixes.

In plain terms
  • Removes numexpr support.
  • Removes the deprecated dask._compatibility.entry_points.
  • Changes the `zarr_read_kwargs` argument to a `mode` argument.
  • Dask now has early support for the free-threaded build of Python 3.14.
Must act Top 5 of 7 that need code or config changes

What you must change

Breaking3
Removes numexpr support.
numexpr
2026.6.0GitHub ReleaseSummarizedView source
Removes the deprecated dask._compatibility.entry_points.
dask._compatibility.entry_points
2026.1.0GitHub ReleaseSummarizedView source
Changes the zarr_read_kwargs argument to a mode argument.
zarr_read_kwargsmode
2026.1.2GitHub ReleaseSummarizedView source
Security2
Reverts the change that enabled Jinja2 autoescape to prevent XSS in HTML reprs.
Jinja2autoescape
2026.6.0GitHub ReleaseSummarizedView source
Enables Jinja2 autoescape to prevent XSS in HTML reprs.
Jinja2autoescape
2026.6.0GitHub ReleaseSummarizedView source
Platform1
Adds preliminary Python 3.14t support.
2026.3.0GitHub ReleaseSummarizedView source

Deprecations to plan for

Deprecated1
Notes that the inplace keyword in DataFrame.drop is deprecated in Pandas 3.1.
DataFrame.dropinplace
2026.6.0GitHub ReleaseSummarizedView source

API changes

Generated by DepDelta from published Python source. Not mixed with prose notes.

  • changed: function array_expr_enabled() → function array_expr_enabled() -> builtins.bool
  • changed: function to_zarr(arr, url, component=…, storage_options=…, region=…, compute=…, return_stored=…, zarr_array_kwargs=…, zarr_read_kwargs=…, **kwargs) → function to_zarr(arr, url, component=…, storage_options=…, region=…, compute=…, return_stored=…, mode=…, **zarr_array_kwargs)
  • changed: function svd(a, coerce_signs=…) → function svd(a, coerce_signs=…, full_matrices=…)
  • removed: variable numbagg
  • added: variable HAS_NUMBAGG
  • changed: function refresh(config: dict | None=…, defaults: list[Mapping]=…, **kwargs) -> None → function refresh(config: dict | None=…, defaults: list[Mapping]=…, paths: list[str]=…, env: Mapping[str, str] | None=…) -> None
  • changed: function rename(deprecations: Mapping[str, str | None]=…, config: dict=…) -> None → function rename(deprecations: Mapping[str, str | None]=…, config: dict=…) -> dict[str, str | None]
  • added: function xfail(request: pytest.FixtureRequest)
  • removed: variable HAS_PYARROW
  • changed: function meta_warning(df, method=…) → function meta_warning(df: Any, method: str) -> str
  • removed: function drop_by_shallow_copy(df, columns, errors=…)
  • changed: function cpu_count() → function cpu_count() -> int
  • added: class ContextAwareThreadPoolExecutor(ThreadPoolExecutor)
  • added: method ContextAwareThreadPoolExecutor.submit(fn, /, *args, **kwargs)