narwhals

Changes

💥 Breaking changes

• feat!: remove maintain_order from LazyFrame.unique, as it was never supported (#2247)

✨ Enhancements

• feat: add cum_min to PySpark (#2267)
• feat: make order_by in Expr.over public (#2271)
• feat: Add support for full join (#2126)
• enh: Support diff, shift, is_first_distinct, is_last_distinct for PySpark, SQLFrame, and DuckDB (#2270)
• feat: add cum_max for PySpark (#2265)
• feat: support window operations for DuckDB (#2263)
• feat: track whether expressions are multi-output, unify error messages (#2246)
• feat: Enable spark-like backends in scan_parquet (#2235)
• feat: Binary dtype support (#2243)
• feat!: remove maintain_order from LazyFrame.unique, as it was never supported (#2247)

🐞 Bug fixes

• fix: safety assertion was being thrown for Expr.is_last_distinct with over and _order_by by pandas and pyarrow (#2269)
• chore(typing): add missing _FullContext to CompliantDataFrame (#2251)

📖 Documentation

• feat: make order_by in Expr.over public (#2271)
• doc: fix docs CI (#2259)
• docs: fix docs CI dependencies (#2260)

🛠️ Other improvements

• ci: skip test_join_duplicate_column_names on old polars (#2277)
• ci: Get CI green for cuDF tests (#2276)
• chore: Clean up _arrow (#2275)
• chore(typing): Move DepthTracking* typing (#2272)
• chore: Track ExpansionKind in ExprMetadata (#2266)
• feat: Adds Compliant(When|Then) (#2261)
• chore: Add CompliantExpr._with_metadata (#2262)
• feat: Adds CompliantGroupBy (#2252)
• ci: skip Polars nightly if notebook is still running (#2257)
• chore: Spec CompliantSeries (#2236)
• chore: downstream CI cleanup (validoopsie and pointblank) and re-order dependency groups (#2255)
• chore: Add CompliantDataFrame.native (#2249)
• chore: Spec CompliantLazyFrame (#2232)
• test: Scope dataframe library imports to fixtures in conftest.py (#2241)
• chore: use is_scalar_like in apply_n_ary_operation (#2242)
• chore: move dev dependencies to dependency-groups (#2238)
• test: Fix yet more tests to work without polars (pandas, pyarrow) (#2250)
• test: Fix more tests to work without polars (and pandas, pyarrow) (#2245)
• Fix noxfile.py syntax for nox-2025.2.9 (#2240)

Thank you to all our contributors for making this release possible! @EdAbati, @FBruzzesi, @MarcoGorelli, @dangotbanned, @marvinl803, @mgorny, @raisadz and @tylerriccio33

Changes

💥 Breaking changes

• feat!: require at least one expression be passed to lazyframe select and lazyframe.with_columns, remove lazyframe.clone (#2206)

⚠️ Deprecations

• docs: replace remaining native_namespace refs (#2221)
• chore: Deprecate native_namespace in favour of backend in from_arrow (#2219)
• chore: Deprecate native_namespace in favour of backend in scan_csv (#2218)
• chore: Deprecate native_namespace in favour of backend in scan_parquet (#2217)
• chore: Deprecate native_namespace in favour of backend in read_parquet (#2216)
• chore: Deprecate native_namespace in favour of backend in from_numpy (#2215)
• chore: Deprecate native_namespace in favour of backend in new_series (#2212)

✨ Enhancements

• feat: Raise uniform error when passing Series of different lengths to with_columns (#2224)
• enh: Adds support for Time datatype (#2113)
• Implement fill_null for spark-like (#2220)
• feat!: remove require at least one expression be passed to lazyframe select and lazyframe.with_columns, remove lazyframe.clone (#2206)

🐞 Bug fixes

• patch: xfail cast_test.py::test_cast_time for cudf (#2231)
• fix(typing): Match lazy @overload in nw.from_native(...) (#2211)
• fix: don't downcast large_string to string unnecessarily in concat_str for PyArrow (#2176)
• refactor: adds _compliant sub-package (#2149)

📖 Documentation

• docs: replace remaining native_namespace refs (#2221)
• docs: add metalearners to used by/ecosystem (#2208)
• docs: add Bokeh to "used by" (#2198)

🛠️ Other improvements

• test: unify tests in dtypes_tests.py (#2144)
• chore(typing): Resolve sqlframe>=3.24.0 errors (#2187)
• chore: pin typing dependnecies (#2233)
• test: Run TPCH with SQLFrame (#2228)
• chore: Spec CompliantDataFrame (#2223)
• refactor: Add CompliantSeries.from_numpy (#2196)
• chore(typing): Fill out CompliantNamespace protocol (#2202)
• chore: minor duckdb cleanup (#2204)
• test: track and enforce sqlframe test coverage (#2200)
• chore(ruff): add format to builtins-ignorelist (#2199)
• chore(typing): Use SQLFrame instead of PySpark for _spark_like internally (#2190)
• refactor: adds _compliant sub-package (#2149)
• ci: unxfail sqlframe (#2186)
• ci: update bump_version.py so it correctly pushes to stable branch (#2183)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @FBruzzesi, @MarcoGorelli, @dangotbanned, @lucas-nelson-uiuc, @luke396 and @marvinl803

Changes

💥 Breaking changes

• fix!: align Series.filter param name with Polars (#2164)

🚀 Performance improvements

• perf: simplify pandas scatter (#2154)

✨ Enhancements

• feat: support passing multiple arguments positionally to get_native_namespace (#2178)
• feat: Added struct namespace with field method. (#2146)
• feat: add rolling_sum for sqlframe and pyspark (#2168)
• feat: support cum_sum for lazy backends (#2132)

🐞 Bug fixes

• fix: parse timezone-aware datetime strings as UTC consistently across backends (#2166)
• fix!: align Series.filter param name with Polars (#2164)
• fix(typing): Narrow NativeSeries Protocol (#2159)

📖 Documentation

• docs(ruff): Improve config docs (#2173)
• docs: even more concise docstrings (#2068) (#2163)

🛠️ Other improvements

• test: fixup cudf failures (#2181)
• chore(typing): remove unused pandas-stubs ignores (#2172)
• chore: remove unnecessary testing logic for pyspark (#2170)
• refactor: adds _Stores(Native|Compliant) (#2130)
• chore(typing): SparkLikeExpr properties (#2152)
• ci: Add darts to downstream tests (#2118)
• ci: fix plotly requirement files in downstream tests (#2161)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MarcoGorelli, @dangotbanned, @osoucy, @skritsotalakis and @DeaMariaLeon

Changes

• feat: Spec'd-out CompliantExpr (#2119)
• [pre-commit.ci] pre-commit autoupdate (#2139)

✨ Enhancements

• feat: support reversed cumulative expressions in over for pandas-like (#2141)
• feat: support anonymous expressions in over context for pandas-like, support over for multi-partitioned dask (#2138)

🐞 Bug fixes

• fix: pyarrow scatter was returning incorrect results when indices were not sorted (#2156)
• ci: fix doc build for win32 (#2147)
• fix: pass ddof down in std().over() for pandas-like (#2143)

📖 Documentation

• docs: Add info in nw.when regarding chained statements (#2151)
• ci: fix doc build for win32 (#2147)

📦 Build system

• ci: fix doc build for win32 (#2147)

🛠️ Other improvements

• docs: remove non-existent "args" (#2153)
• ci: get cuDF test suite green again (#2145)
• refactor: Use chain.from_iterable more (#2140)
• refactor: allow callable in CompliantExpr.from_column_names to reuse for CompliantNamespace.<all|col|exclude> (#2134)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MarcoGorelli, @dangotbanned, @pre-commit-ci[bot] and pre-commit-ci[bot]

Changes

• test: enhance xfail markers with specific reasons in timestamp_test.py (#2065)
• refactor: Generic CompliantSelector (#2064)
• docs: more concise docstrings(#2068) (#2117)
• chore: deprecate LazyFrame.gather_every in main namespace (but maintain it in stable.v1) (#2103)
• test(typing): fix DataFrame.sort_values overload match (#2109)

🚀 Performance improvements

• perf: Avoid redefining lambda in *Namespace.all (#2102)
• perf: avoid full broadcast for otherwise_value in when/then/otherwise (#2098)
• perf: use cached property for schema/columns lazy frames (#2093)
• perf: only validate duplicate column names when collecting for duckdb/pyspark/dask (#2092)

✨ Enhancements

• feat: Adds nw.exclude (#2122)
• feat: Add support for pandas-like .diff().over(group) (#2128)
• fill_null with expression (#2106)
• feat: Adds DataFrame.iter_columns (#2104)
• feat: str.split (#1932) (#2054)
• feat(typing): Backport generic Series to v1 (#2110)
• feat: align over signature with Polars (#2096)
• chore: Finalize support for SQLFrame (#2038)

🐞 Bug fixes

• fix(typing): Include Expr in fill_null signature (#2129)
• fix: Allow exprs in .filter constraints (#2114)
• fix: pyspark unpivot with index None (#2090)

📖 Documentation

• docs: Add API example docs for Field dtype (#2131)
• docs: add Implementation page (#2107)
• docs: concise_docstrings (#2068) (#2086)

🛠️ Other improvements

• refactor: Simplify ArrowGroupBy.__iter__ (#2133)
• chore: simplify pandas-like over (#2127)
• chore(typing): Generic CompliantDataFrame (#2115)
• chore(typing): enable typing checks for pyspark (#2051)
• chore(typing): remove unused, ignored TypeVar (#2108)
• chore(typing): fix group_by (#2105)
• perf: Avoid redefining lambda in *Namespace.all (#2102)
• chore: rename changes_length to filtration (#2094)
• chore: add has_open_windows to ExprMetadata instead of is_order_dependent (#2078)
• chore(typing): Resolve _dask errors (#2087)
• refactor(typing): Use a forward ref for IntoCompliantExpr (#2088)
• chore(typing): Fix (Pandas|Arrow)(When|Then) (#2089)
• refactor(ruff): enable some preview rules (#2042)

Thank you to all our contributors for making this release possible! @EdAbati, @FBruzzesi, @MarcoGorelli, @dangotbanned, @janpipek, @luke396, @skritsotalakis and @thomasjpfan

Changes

🚀 Performance improvements

• perf: avoid repeating constant iteration (#2079)
• chore(perf): move combine_evaluate_output_names check (#2040)
• perf: Avoid double-parsing in concat_str (#2039)

✨ Enhancements

• enh: Deprecate native_namespace in favour of backend in read_csv (#2063)

🐞 Bug fixes

• chore: remove unused kwargs in _pandas_like and _arrow (#2069)
• fix: disallow non-elementary operations in .over for pandas (#2062)
• fix: Type conversion from Spark Struct to narwhals Struct. (#2037)
• fix: Use metaclass for safe DType attr access (#2025)

📖 Documentation

• docs: remove griffe fork to build docs (#2080)
• chore: remove unused kwargs in _pandas_like and _arrow (#2069)
• docs: Add standardized test failure patterns to CONTRIBUTING.md (#2052)

🛠️ Other improvements

• chore(typing): Remove unused pyright ignores (#2076)
• chore(typing): Re-enable mypy strict (#2077)
• refactor: Use real class for ExprMetadata (#2074)
• chore: use combine_metadata more (#2073)
• chore: use default_metadata (#2072)
• chore: use more expression metadata functions (#2071)
• chore: use Expr._from_callable (#2070)
• chore: remove unused kwargs in _pandas_like and _arrow (#2069)
• chore: use change_metadata_kind (#2067)
• chore(typing): add CompliantSeries.dtype (#2066)
• refactor(typing): make _evaluate_output_names generic (#2053)
• refactor(typing): Avoid more ignore[attr-defined] on Datetime (#2049)
• chore: remove more kwargs from dask, dont treat ddof as expressifiable arg (#2059)
• chore(typing): add CompliantExpr._version (#2060)
• chore: remove kwargs from selector function (#2058)
• refactor: deduplicate *.selectors (#2057)
• ci(typing): add pyright to ci (#2035)
• chore: refactor EVERYTHING to do with broadcasting (#2005)
• ci: dont pass filenames to mypy in pre-commit (#2046)
• chore: add colors in CI and run marimo only in selected python version (#2043)
• chore(perf): move combine_evaluate_output_names check (#2040)
• chore: refactor right-hand-side arithmetic to not use lit (#2034)

Thank you to all our contributors for making this release possible! @EdAbati, @FBruzzesi, @MarcoGorelli, @dangotbanned, @luke396, @osoucy and @raisadz

Changes

🐞 Bug fixes

• fix: PySpark was raising during collect when it contained no rows and a void dtype column (#2032)
• fix: is_in was raising when used with empty list (#2031)

🛠️ Other improvements

• chore: fix pandas-stubs issues (#2008)
• fix(typing): Resolve all mypy & pyright errors for _arrow (#2007)

Thank you to all our contributors for making this release possible! @EdAbati, @MarcoGorelli and @dangotbanned

Changes

🚀 Performance improvements

• perf: Avoid evaluating expr twice in mean_horizontal (#2028)
• perf: avoid try/except in _hasattr_static (#2014)
• perf: fast path for pandas-like group by with single expression and avoid resetting index if not necessary (#2010)

✨ Enhancements

• feat: support casting to and from spark-like structs (#1991)
• feat: Add support for series[other_series] (#2013)

🐞 Bug fixes

• fix: adjust duckdb population skewness by correction factor to get sample skewness (#2021)
• fix: sqlframe false positives in compliance checks (#2011)

🛠️ Other improvements

• chore(typing): Upgrade TypeGuard -> TypeIs (#2027)
• refactor: remove Array shape default (#2026)
• refactor: reuse Schema.to_<backend>() in from_numpy (#2024)
• chore(typing): relax from_numpy to use Mapping, Sequence (#2023)
• chore(typing): relax from_dict to use Mapping (#2022)
• tests: Allow to run pytest narwhals --doctest-modules without ibis (#2019)
• ci(typing): use tool.mypy.pretty = true (#2018)
• fix: various typos (#2016)
• ci: update pointblank deps to test downstream in ci (#2015)
• chore(typing): resolve time unit/zone set invariance (#2012)
• test: add test for sum_horizontal with column transformation (#2009)
• chore: simplify when implementation (#2003)
• ci: remove pyspark from Windows CI (#1998)
• chore(typing): add overloads to isinstance_or_issubclass (#1997)
• chore: parse series and ndarrays as expressions (#1994)
• chore(typing): remove unused ignores (#1979)
• fix: use mypy pre-commit in local environment (#1966)
• chore: clean up flatten and extract (#1995)
• chore: determine whether to use compliant_frame.aggregate at narwhals level (#1993)
• chore: Refactor filter (#1992)
• chore: alias named expressions at narwhals level (#1990)
• chore: Expr metadata refactor (#1986)
• chore: use toArrow in PySpark 4.0+ (#1987)
• tests: skip cudf hist tests (#1988)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @EdAbati, @FBruzzesi, @MarcoGorelli, @anopsy, @camriddell, @dangotbanned and @raisadz

Changes

• ci: validate tpch query results (#1971)
• ci: add pointblank in downstream test (#1956)
• ci: add Validoopsie to downstream tests (#1973)
• chore(typing): fix _from_compliant_dataframe type ignore (#1969)
• fix(typing): resolve remaining utils.py warnings (#1968)
• test: Add new_streaming (#1898)
• fix(typing): add missing Series._level annotation (#1967)
• ci: fix plotly tests location (#1965)

✨ Enhancements

• feat: Series.hist (#1859)
• feat: catch Polars exceptions, unify exception raising more (#1918)
• feat: improving array casting (#1865)
• feat: add datetime selector (#1822)
• feat: add Schema.to_(arrow|pandas|polars) (#1924)

🐞 Bug fixes

• fix: implement casting to Int128 (#1984)
• fix: fix type __get_item__ (#1958)
• fix(typing): Backport from_native overload fixes to v1 (#1953)
• fix(typing): Backport use IntoDataFrameT in to_native to v1 (#1954)

📖 Documentation

• docs: add Schema.to_(arrow|pandas|polars) (#1980)
• RFC, chore: shrink series module docstrings (#1976)
• chore: shrink functions module docstrings (#1974)
• docs: add docstring guidelines to CONTRIBUTING.md (#1957)

🛠️ Other improvements

• chore: minor series.py fixes, dedupe a little (#1982)
• chore(typing): resolve _arrow.dataframe warnings (#1983)
• chore(typing): add np.ndarray aliases (#1977)
• chore: widen parse_version to accept packages (#1975)
• RFC, chore: shrink series module docstrings (#1976)
• chore: shrink functions module docstrings (#1974)
• chore(typing): fix remaining functions, series warnings (#1970)
• fix: missing __getitem__ type fixes (#1963)
• chore(typing): add guards is_compliant_* (+ friends) (#1955)
• chore: reduce duplication, implement more methods at the Narwhals level rather than at the compliant level, fix CI from Ibis release (#1960)

Thank you to all our contributors for making this release possible! @EdAbati, @FBruzzesi, @IsaiasGutierrezCruz, @MarcoGorelli, @akmalsoliev, @camriddell, @dangotbanned and @luke396

Changes

✨ Enhancements

• feat: restore support of is_duplicated for pyarrow (#1951)

🐞 Bug fixes

• fix: cuDF to_numpy tests were failing (#1949)
• fix: raise NotImplementedError for DataFramte.is_duplicated for PyArrow when there is more than one column (#1948)
• fix: wrong argument name suggested in from_dict deprecation (#1946)

Thank you to all our contributors for making this release possible! @FBruzzesi and @MarcoGorelli

Changes

🚀 Performance improvements

• perf: use fastpath in DataFrame.to_numpy for pandas, improve performance for DataFrame.schema for pandas, use fewer values to sniff dtype for pandas objects (#1929)

✨ Enhancements

• enh: Deprecate native_namespace in favour of backend in from_dict (#1931)
• feat: validate duplicate column names in pyarrow and duckdb (#1815)

🐞 Bug fixes

• fix: Always collect Iterator[IntoExpr] in utils.flatten (#1934)

📖 Documentation

• docs: Make DataFrame and LazyFrame docstrings shorter and more concise (#1939)
• chore: fix a couple of typos 🙃 (#1921)

🛠️ Other improvements

• ci: fix plotly downstream test (#1938)
• fix(typing): Use IntoDataFrameT in to_native (#1936)
• fix(typing): Use correct TypeVar/Union pairs in from_native (#1935)
• fix(typing): Ensure LazyFrame has a unique TypeVar (#1930)
• refactor: refactoring following refurb and perflint suggestions (#1925)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @FBruzzesi, @IsaiasGutierrezCruz, @MarcoGorelli, @dangotbanned, and @raisadz

Changes

✨ Enhancements

• feat: add LazyFrame.explode for duckdb (#1891)
• feat: add duckdb .str.to_datetime (#1916)
• feat: enable ModuleType and str for backend in .lazy() method (#1914)
• feat: add LazyFrame.unpivot for spark and duckdb (#1890)
• feat: LazyFrame.collect with backend and **kwargs (#1734)
• feat: spark date and datetimedata types, allow for dtype in nw.lit for pyspark (#1913)

📖 Documentation

• docs: Minor rendering fixups, inherit all exceptions from NarwhalsError (#1919)
• docs: shorten docstring examples in narwhals/expr.py (#1915)

🛠️ Other improvements

• chore: simplify duckdb group-by (#1911)

Thank you to all our contributors for making this release possible! @FBruzzesi, @osoucy , and @MarcoGorelli

Changes

✨ Enhancements

• feat: Add backend argument to lazy (#1895)
• feat: add dtype.is_<x> methods (#1906)
• feat: spark like .list namespace and lazyframe .explode (#1887)

🐞 Bug fixes

• fix: pandas was raising when index name and column names overlapped in groupby (#1908)
• docs: update selectors docstrings (#1905)

📖 Documentation

• docs: improve search in docs (#1909)
• docs: update selectors docstrings (#1905)
• docs(ecosystem.md): add pointblank to Used By section (#1904)
• docs(README.md): add mention of pointblank project (#1903)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @FBruzzesi, @MarcoGorelli, @raisadz and @rich-iannone

Changes

• chore: fix pyspark/sqlframe typo (#1884)

✨ Enhancements

• feat: start (silently) adding support for SQLFrame (#1883)
• feat: move robust literal handling for PySpark (#1880)
• feat: add selectors.matches (#1882)
• feat: support more scalar operations for duckdb, Increase width for ipython (#1877)

🛠️ Other improvements

• chore: add "typing :: typed" classifier to pyproject.toml (#1881)
• chore: use positional-only arg in evaluate_into_exprs (#1879)
• chore: remove unused _depth from DuckDBExpr and SparkLikeExpr (#1878)
• chore: refactor name namespaces to lower code duplication (#1876)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MarcoGorelli and @raisadz

Changes

• docs: fix api completness generation in CI (#1863)

✨ Enhancements

• feat: pyspark and duckdb selectors (#1853)
• feat: completely refactor alias tracking and support nw.all, nw.nth, and selectors across the API (#1866)
• feat: raise if expression changes length in .over() (#1867)
• feat: add DuckDB join_asof (#1860)
• feat: allow any ddof value for duckdb var and std (#1858)
• feat: add DuckDB Struct dtype (#1851)
• feat: add DuckDB list.len method (#1850)
• feat: enforce that group-by aggregations actually aggregate (#1844)
• feat: add mean_horizontal for DuckDB (#1846)

🐞 Bug fixes

• fix: support various reductions in pyspark (#1870)
• fix: address & / | operator errors for PySpark / chore: use F.lit in maybe_evaluate for pyspark, like we do for duckdb (#1872)
• fix: allow any name in expression parsing (#1871)
• fix: nw.selectors.by_dtype(nw.Datetime) was excluding tz-aware for Polars (#1869)
• patch: fix duckdb narwhals_to_native_dtype for Int32 and UInt32 (#1857)
• fix: raise on selectors addition (#1854)
• fix: rename ewm and rolling min_periods to min_samples for polars 1.21 compatibility (#1864)
• fix: address & / | operator errors for PySpark / chore: use F.lit in maybe_evaluate for pyspark, like we do for duckdb (#1872)

🛠️ Other improvements

• test: skip Ibis tests if deps aren't installed, set defaults in PandasLikeSeries.__array__ to fix cuDF is_in (#1862)
• chore: parse strings as columns names at narwhals-level (#1856)
• chore: some typing and default removal (#1852)
• chore: validate expressions shape at narwhals/expr.py level (#1845)
• chore: simplify internal renaming in _pandas_like/series.py (#1847)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MarcoGorelli and @raisadz

Changes

• ci: downstream test for formulaic (#1817)
• docs: Removed class, added polars to the table, str.head and str.tail are a… (#1801)

✨ Enhancements

• feat: deprecate maintain_order in Expr.unique and LazyFrame.tail (but keep around in stable.v1) (#1839)
• feat: track whether expressions change length but don't aggregate, and only allow length-changing expressions if they're followed by aggregations in the lazy API (#1828)
• feat: add DuckDB: nw.nth, nw.sum_horizontal, nw.concat_str, group_by with drop_null_keys (#1832)
• feat: remove eager_or_interchange from from_native in main namespace, switch Ibis' support from interchange to lazy in main namespace (but preserve status-quo in stable.v1) (#1829)
• feat: add is_nan and is_finite for duckdb, is_nan for pyspark (#1825)
• chore: rename time zone tests (#1830)
• feat: Deprecate Expr.arg_true (but keep Series.arg_true, and keep both available in stable.v1) (#1827)
• feat: add Expr.dt methods to PySpark (#1835)
• feat: add pyspark str namespace to_datetime (#1826)
• feat: add total duration methods for DuckDB (#1831)
• feat: add .over method for SparkLikeExpr (#1808)
• feat: pyspark group by n_unique and no aggregation (#1819)
• feat: Disallow order-dependent expressions from being passed to nw.LazyFrame (#1806)
• feat: add SparkExpr.cast for basic types (#1812)
• feat: pyspark and duckdb Expr.name namespace (#1809)
• feat: add AnonymousExprError (#1816)
• feat: deprecate Expr.head, Expr.tail, Expr.sort, Expr.gather_every, Expr.sample (but keep them in stable.v1) (#1791)
• feat: add duckdb dataframe drop_nulls (#1811)
• feat: add DataFrame and Series to_polars (#1803)
• feat: add support for SparkLikeNamespace.when (#1805)

🐞 Bug fixes

• fix: when/then/otherwise output name was not consistent across backends (#1833)
• fix: raise on shape mismatch in filter (#1814)
• patch: fix when-then double lit case (#1810)

🛠️ Other improvements

• chore: simplify imports (#1838)
• test: fixup test suite for cudf.pandas (#1837)
• test: decouple pyspark constructor from pandas (#1834)
• chore: refactor pandas-like narwhals_to_native_dtype (#1824)
• chore: refactor isinstance_or_issubclass to allow tuples (#1820)

Thank you to all our contributors for making this release possible! @Dhanunjaya-Elluri, @EdAbati, @FBruzzesi, @MarcoGorelli, @luke396, @marvinl803 and @raisadz

Changes

✨ Enhancements

• feat: improve error message on check_columns_exist (#1799)
• feat: DuckDB date (#1798)
• feat: DuckDB dt.to_string (#1794)
• feat: DuckDB dt.ordinal_day (#1796)
• feat: DuckDB dt.weekday (#1795)
• feat: add extra overload for concat for when it is not known statically whether inputs are eager or lazy (#1783)
• feat: duckdb replace_all with regex (#1784)
• feat: implement anti-join, str.len_chars, and null_count for DuckDB (#1777)
• feat: SparkLikeNamespace methods (#1779)
• feat: add SparkLikeStrNamespace methods (#1781)
• feat: add all, any and null_count Spark Expressions (#1724)
• feat: add missing dunder methods in SparkLikeExpr and SparkLikeNamespace.lit (#1708)
• feat: implement cross-join for duckdb (#1773)
• feat: semi-join for duckdb (#1767)
• feat: implement n_unique for DuckDB (#1762)
• feat: implement when/then/otherwise for DuckDB (#1759)
• feat: add few missing SparkLikeExpr methods (#1721)

🐞 Bug fixes

• fix: to_py_scalar was raising for decimals (#1800)
• fix: fix skew for duckdb with fewer than 3 elements (#1785)
• fix: duckdb join was failing if column names contained spaces (#1775)
• fix: parse_version was not parsing duckdb pre-preleases correctly (#1763)
• fix: update Spark min version in utils.py (#1760)

📖 Documentation

• docs: remove docstring examples from narwhals/stable/v1 (#1797)
• docs: Restore API complentess page (#1788)
• docs: Increased width for content (#1769)
• docs: fix is_between type hint in signature (#1766)

🛠️ Other improvements

• chore: factor out check_columns_exist (#1792)
• chore: split namespaces out from expr and series (#1782)
• chore: move pyspark tests into main test suite (#1761)
• chore: dask nightly (#1768)
• chore: Remove some unnecessary trailing commas (#1757)
• chore: validate predicates in nw.when one level higher (#1756)
• chore: remove some expr._kwargs defaults (#1747)
• chore: Filter left join warning (#1753)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @Dhanunjaya-Elluri, @EdAbati, @FBruzzesi, @MarcoGorelli, @camriddell, @lucas-nelson-uiuc, @marvinl803 and @raisadz

Changes

✨ Enhancements

• feat: add Series|Expr.rank (#1342)

🐞 Bug fixes

• fix: fix license classifier (#1751)

📖 Documentation

• docs: update api-completeness with duckdb (#1740)

🛠️ Other improvements

• test: remove cudf from tests/expr_and_series/replace_time_zone_test.py::test_replace_time_zone_none[cudf] (#1748)
• chore: filter old pyarrow/pandas warnings we cant do anything about (#1746)
• [pre-commit.ci] pre-commit autoupdate (#1741)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @Dhanunjaya-Elluri, @EdAbati, @FBruzzesi, @MarcoGorelli, @anopsy

Changes

• fix: always pass Column to pandas.spark.functions (#1712)
• tests: add 'core' to nox dependencies (#1713)
• feat: add weekday to Pandas, Dask and Arrow dt namespaces (#1680)
• Link to Avoiding the UserWarning error while using Pandas group_by page when warning is thrown (#1683)

✨ Enhancements

• feat: Implement partial "lazy" support for DuckDB (even with this PR, DuckDB support is work-in-progress!) (#1725)
• feat: validate library minimum version in compliant objects (#1727)
• feat: enable to cast to Struct dtype (#1717)
• feat: Expressify clip arguments (#1709)
• feat: add is_nan expression & series method (#1625)
• feat: show native object (if possible) in repr (#1702)
• feat: add sum and sum_horizontal for SparkLike (#1693)
• feat: add some SparkLikeLazyFrame methods (#1633)

🐞 Bug fixes

• fix: fix broken link from warning (#1732)
• fix: nw.lit(date, dtype=nw.Date), loosen Dask minimum back to 2024.8 (#1730)
• chore: simplify definitions of rhs expressions, bump dask minimum to 2024.10 (#1720)
• fix: dask group by with kwargs (#1676)
• fix: DaskLazyFrame and ArrowDataFrame .with_row_index column order (#1706)
• fix: scalar reductions on empty inputs (#1715)
• fix: remove maintain_order from LazyFrame.unique (#1687)
• fix: modin dtype interoperability (#1692)
• fix: Dask was raising for scalar vs Series binary operations (#1684)
• fix: casting to List for cudf (#1686)

📖 Documentation

• docs: Expr method' docstrings (#1733)
• docs: let nw.LazyFrame docstrings examples run on Polars and Dask only (#1700)
• docs: update null handling page with is_nan reference (#1716)
• docs: add spark-like in backend completeness (#1710)
• docs: Series method' docstrings (#1699)
• docs: DataFrame method' docstrings (#1688)

🛠️ Other improvements

• test: catch some cudf failures (#1735)
• chore: refactor root_names and output_names tracking (#1731)
• tests: remove unnecessary pytest filterwarnings (#1691)
• chore: remove some lazy tests (#1719)
• fix: Catch cuDF warning for concat with empty elements, and ComputeError for non-float is_nan (#1718)
• chore: use None instead of float('nan') to check for null values in tests (#1697)
• test: allow to run tests for Polars[gpu] (#1698)
• test: Make constructors configurable when running tests (#1694)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @EdAbati, @FBruzzesi, @MarcoGorelli, @camriddell, @dependabot, @dependabot[bot], @lucas-nelson-uiuc and @lucianosrp

Changes

🐞 Bug fixes

• fix: is_duplicated was returning wrong-length result for PyArrow and Dask (#1679)

Thank you to all our contributors for making this release possible! @DeaMariaLeon and @MarcoGorelli

Changes

🚀 Performance improvements

• chore, perf: arrow cat namespace (#1663)

✨ Enhancements

• feat: raise informative error message when grouping by expressions (#1677)
• feat: expressify lower_bound and upper_bound in is_between (#1672)
• feat: Relax cuDF minimum version to 24.10.0 (#1649)
• Feat/add collect schema to interchange dfs (#1646)
• feat: Add support for .shift(n).over('col') for pandas-like DataFrames (#1627)

🐞 Bug fixes

• fix: pyspark group by with kwargs (#1665)

📖 Documentation

• docs: use IntoDataFrameT/IntoFrameT in docs (#1664)
• docs: Add return type description for more series namespace (#1675)
• docs: Add more return type descriptions for Series (#1674)
• docs: add return type descriptions for cum_count, cum_max, cum_min, cum_prod, cum_sum in Series (#1673)
• docs: add return type descriptions to Series arg_true, cast, clip, count (#1667)
• docs: Add return descriptions to Series's abs, alias, all, any methods for clarity (#1653)

🛠️ Other improvements

• chore: pyarrow do not combine chunks (#1662)
• chore: flatten subset in unique one level up (#1671)
• test: add test for nw.Expr|Series.str.contains with literal=True (#1670)
• test: check keys are in the same order in assert_equal_data (#1668)
• ci: improve publish ci to use env vars (#1652)
• chore: use __narwhals_namespace__ more (#1658)
• ci: split dev from extra (#1655)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @ClaudioSalvatoreArcidiacono, @Dhanunjaya-Elluri, @FBruzzesi, @MarcoGorelli and @luke396

Changes

🚀 Performance improvements

• perf: avoid merge in pandas groupby (#1638)

✨ Enhancements

• feat: add collect schema to interchange dfs (#1646)
• feat: support std and var with ddof !=1 in pandas-like group by (#1645)
• feat: add DataFrame and LazyFrame explode method (#1542)

🐞 Bug fixes

• fix: Make compatible with latest cuDF release (#1640)

📖 Documentation

• docs: Note that maintains_order has no effect (but keep around for backwards-compatibility) (#1643)
• docs: fix typing example (#1637)
• docs: Use relative paths (#1636)
• docs: Remove warnings when building docs (#1631)
• docs: returns and arguments documentation for lazyframe class (#1630)
• docs: null handling (#1624)

🛠️ Other improvements

• chore: Update pyproject.toml (#1642)
• chore: register kwargs for CompliantExpr (#1614)
• fix: exclude protocols in api completeness script (#1635)
• test: not-so-old-versions, pretty-old-versions, nightlies (#1626)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @Dhanunjaya-Elluri, @FBruzzesi, @MarcoGorelli, @marenwestermann and @marvinl803

Changes

🚀 Performance improvements

• perf: streamline pyarrow and pandas group by agg (#1621)

✨ Enhancements

• feat: add Series|Expr.rolling_var and Series|Expr.rolling_std (#1451)
• fix: Add support for pandas cumulative operations with over (#1565)
• feat: add var to Expr and Series (#1603)

🐞 Bug fixes

• fix: nw.col('a').is_in(df['b']) was raising (#1616)
• fix: pandas and pyarrow were raising for repeated aggregations involving the same root name and the same aggregation function (#1613)

📖 Documentation

• docs: more args and returns documentation in DataFrame class (#1600)
• docs: make docstring examples function names more descriptive in functions.py (#1602)
• docs: improve DataFrame, LazyFrame and Series docstrings (#1622)

🛠️ Other improvements

• ci: unpin scikit-learn in scikit-lego test (#1611)
• ci: Speed up tests in local development (#1608)
• test: downstream test for tea-tasting (#1605)
• test: minimum versions (#1617)

Thank you to all our contributors for making this release possible! @ClaudioSalvatoreArcidiacono, @DeaMariaLeon, @Dhanunjaya-Elluri, @FBruzzesi, @Machele-codez, @MarcoGorelli, @e10v, @marenwestermann and @skritsotalakis

Changes

✨ Enhancements

• feat: add Series and Expr arg_min & arg_max (#1529)

🐞 Bug fixes

• fix: allow np.scalar to be used in Series.getitem (#1515)
• fix: fix pyarrow len aggregation behaviour for depth-1 exprs (#1589)

📖 Documentation

• docs: document arguments and returns in DataFrame class (#1594)
• docs: add tea-tasting to ecosystem (#1593)
• docs: Correct small error in IntoDataFrameT and IntoSeriesT (#1591)
• docs: add docstring example for Array in narwhals.dtypes (#1576)

🛠️ Other improvements

• test: xfail cudf argmin / argmax (#1599)
• chore: use set_axis in maybe_set_index (#1598)
• test: fixup more cuDF xfails (#1596)
• test: xfail test_group_by_shift_raises for cudf (#1595)
• chore: Add some Compliant Protocols (#1522)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @Dhanunjaya-Elluri, @MarcoGorelli, @Nathan-Bransby-NMT, @e10v, @marenwestermann and @thevro

Changes

• release: Bump version to 1.18.2 (#1582)

✨ Enhancements

• feat/fix: Improve error message when trying to do unsupported group-by aggregation, and raise instead of giving inconsistent results (#1584)

🛠️ Other improvements

• refactor: PandasLikeExpr repr and init (#1585)

Thank you to all our contributors for making this release possible! @MarcoGorelli, @anopsy and @camriddell

Changes

🚀 Performance improvements

• perf: use lru_cache for native_to_narwhals_dtype (https://github.com/narwhals-dev/narwhals/pull/1564)

✨ Enhancements

• feat: add Decimal dtype (https://github.com/narwhals-dev/narwhals/pull/1571)
• feat: Preserve null values in old Polars versions for ewm_mean (https://github.com/narwhals-dev/narwhals/pull/1574)
• enh: add read_parquet / scan_parquet (https://github.com/narwhals-dev/narwhals/pull/1577)
• feat: add nw.Int128, nw.UInt128 (https://github.com/narwhals-dev/narwhals/pull/1570)
• feat: add **kwargs to read_csv and scan_csv (https://github.com/narwhals-dev/narwhals/pull/1560)
• feat: scan_csv (https://github.com/narwhals-dev/narwhals/pull/1555)

📖 Documentation

• docs:add docstring series getitem (https://github.com/narwhals-dev/narwhals/pull/1559)
• docs: Restore image in docs (https://github.com/narwhals-dev/narwhals/pull/1573)
• docs: add security page to docs, try removing some temporary CI pins (https://github.com/narwhals-dev/narwhals/pull/1568)
• docs: add nw.Expr.list, nw.Series.list to api-reference (https://github.com/narwhals-dev/narwhals/pull/1558)

🛠️ Other improvements

• ci: fixup release process (https://github.com/narwhals-dev/narwhals/pull/1580)
• refactor: use nw.scan_parquet in tpch benchmarks (https://github.com/narwhals-dev/narwhals/pull/1578)
• ci: remove pull_request_target, sanitise github.ref_name (https://github.com/narwhals-dev/narwhals/pull/1567)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @MarcoGorelli, @TimOsahenru, @anopsy and @raisadz

Changes

✨ Enhancements

• feat: read_csv (#1551)
• feat: Adding dataframe estimated size (#1549)
• enh: nw.dependencies.is_narwhals_dataframe / nw.dependencies.is_narwhals_lazyframe / nw.dependencies.is_narwhals_series (#1550)
• feat: ListNamespace.len() for Expr and Series (#1536)

🐞 Bug fixes

• fix: return Python scalar for Series.getitem when pyarrow-backed (like Polars does) (#1554)
• fix: .list namespace should preserve pandas index (#1538)

📖 Documentation

• docs: Add a docstring example in narwhals.dtypes for Datetime (#1544)

🛠️ Other improvements

• ci: temporary pins to get CI green (#1552)
• ci: actually test against pandas nightly, don't filter "copy is deprecated" warnings (#1547)
• chore: avoid unnecessary branching on pl version in pln.mean (#1543)
• ci: fix template name (#1540)
• ci: try using "unrelease" version-template for release drafter (#1539)
• chore: rename series and expr to _compliant for Arrow, Polars, and Dask (#1537)
• ci: actually test against pandas nightly, don't filter "copy is deprecated" warnings (#1547)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @FBruzzesi, @Machele-codez, @MarcoGorelli, @anopsy, @marvinl803, @olp-cs and @raisadz

Changes

✨ Enhancements

• feat: get_native_namespace accepts native objects (#1520)
• feat: Expose nw.Implementation, along with Implementation.is_pandas, Implementation.is_pandas_like, and more (#1531)
• feat: Add minimal PySpark support (#908)

📖 Documentation

• docs: add link to "Generating a new SSH key" (#1530)
• docs: Add narwhals.dtypes.Date docstring example (#1528)
• docs: add example to Unknown dtype (#1521)
• docs: add Duration dtype docstring example (#1516)
• docs: add Enum dtype docstring example (#1514)
• docs: add Object dtype docstring example (#1511)
• docs: add return description for Series.to_list (#1513)
• docs: update contributing guide by adding upstream remote (#1509)

🛠️ Other improvements

• chore: remove many false positives from "ignore banned import" check (#1532)
• chore: rename _expr to _compliant_expr in PandasLikeExpr (#1527)
• chore: overload parse_into_expr (#1519)
• chore: rename _call to _to_compliant_expr in narwhals/expr.py (#1518)
• chore: _pandas_series to _compliant_series in PandasSeries (#1517)
• test: fix nox dependencies (#1497)

Thank you to all our contributors for making this release possible! @BenjaminFraser, @EdAbati, @MarcoGorelli, @camriddell, @chrisschopp, @czhanger, @munsoor, @olp-cs, @ontowhee, @reshamas and @skritsotalakis

Changes

• patch: fix dtypes rendering (#1481)

✨ Enhancements

• feat: raise informative error message for DataFrame.eq/neq (#1494)
• feat: add Series.__contains__ (#1480)
• feat: improve invalid concat error message (#1491)

🐞 Bug fixes

• fix: nw.Series could not be pickled (#1488)
• fix: Set level correctly for lazy-only implementations (#1478)

🛠️ Other improvements

• chore: Remove unused code in pandas like series (#1489)
• chore: rename _change_dtypes to _change_version (#1490)
• chore: rename validate_column_comparand utility functions, add pep740 badge (#1477)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MUKESHRAJMAHENDRAN, @MarcoGorelli

Changes

🐞 Bug fixes

• fix: follow left-hand-rule in index alignment in from_dict (https://github.com/narwhals-dev/narwhals/pull/1475)

🛠️ Other improvements

• chore: polars type hinting (https://github.com/narwhals-dev/narwhals/pull/1467)

Thank you to all our contributors for making this release possible! @FBruzzesi and @MarcoGorelli

Changes

✨ Enhancements

• feat: consistently return Python scalars from Series reductions for PyArrow (#1471)
• feat: make Series generic (#1412)
• feat: create dataframe from 2D numpy array and column names (#1456)
• feat: diagonal concat (#1455)

🐞 Bug fixes

• fix: address lit broadcasting and output name of right arithmetic ops (#1424)

📖 Documentation

• docs: clean lazyframe docstrings (#1466)
• docs: ecosystem and resources (#1468)
• docs: add from/to_native in documentation examples (#1461)

🛠️ Other improvements

• ci: Remove vegafusion from blocking releases, assert current branch in bump_version (#1470)
• chore: PyArrow type hinting (#1465)
• ci: add hierarchicalforecast (#1460)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @EdAbati, @FBruzzesi, @MarcoGorelli and @raisadz

Changes

• BUG: remove cudf from import hooks (#1445)

✨ Enhancements

• feat: allow for "import narwhals.stable.v1.selectors as cs" (#1459)
• feat: raise NotImplementedError for not supported parameters in ewm_mean for cuDF (#1449)
• feat: silently support Fireducks and cudf.pandas (#1439)
• feat: improve concat error message and concat docs (#1435)

🐞 Bug fixes

• fix: Preseve dtypes modules in from dict (#1458)

📖 Documentation

• docs: fixup typing in docstrings (#1452)
• docs: fixup copy button (#1440)
• docs: Add docstrings with examples to types (#1438)

🛠️ Other improvements

• ci: Check docstrings don't have unused imports in docstring examples (#1453)
• ci: get cudf tests passing again (#1446)
• chore: add backend_version parameter to polars dtype translation utility (#1441)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @MarcoGorelli and @raisadz

Changes

✨ Enhancements

• feat: improve error message when casting to invalid type (#1429)
• feat: add support for Series|Expr.skew method (#1173)
• feat: make sure from narwhals.stable.v1.dependencies import works (#1430)
• feat: support constraints in DataFrame.filter (#1417)
• feat: add Expr|Series.rolling_mean method (#1290)

🐞 Bug fixes

• fix: return Python bool instead of pyarrow boolean scalar for Series reductions (#1432)
• fix: correct minimum Polars version for ewm_mean (#1415)

📖 Documentation

• docs: use type hints + from_native/to_native in dataframe.py (#1425)
• docs: explain optional boolean args in from_native (#1421)
• docs: Add documentation for LazyGroupBy (#1423)
• docs: add Plotly to the list of projects using Narwhals in README.md (#1418)
• docs: use type hints + from_native/to_native in dataframe.py (#1411)
• docs: use type hints + from_native/to_native in series.py (#1408)
• docs: add vegafusion to "used by" on readme (#1409)
• docs: use from_native / to_native in series.py (#1406)
• docs: use boolean columns in any/all horizontal docstrings (#1403)
• docs: use from_native / to_native in expr.py (#1404)

🛠️ Other improvements

• ci: temporarily pin olama (#1422)
• fix: correct minimum Polars version for ewm_mean (#1415)

Thank you to all our contributors for making this release possible! @CarloLepelaars, @DeaMariaLeon, @FBruzzesi, @MarcoGorelli, @gvwilson, @marenwestermann and @raisadz

Changes

✨ Enhancements

• feat: Adding ewm_mean (#1298)
• feat: add Series|Expr.rolling_sum method (#1395)

🐞 Bug fixes

• fix: remove rogue "t" in ColumnNotFoundError (#1398)

🛠️ Other improvements

• ci: add plotly to downstream tests (#1376)
• ci: prevent ibis from setting upper bounds (where possible) (#1399)

Thank you to all our contributors for making this release possible! @FBruzzesi, @DeaMariaLeon, and @MarcoGorelli

Changes

✨ Enhancements

• feat: add Series|Expr.is_finite method (#1341)
• feat: make exceptions module public (#1391)
• feat: consistently raise ColumnNotFoundError for missing columns in select and drop (#1389)
• feat: add Series|Expr.cum_prod method, add reverse kw in cum_sum method (#1386)
• feat: add Series|Expr cum_min and cum_max methods (#1384)
• feat: Improve error message when comparing Series with list literal, or when using multi-output expressions in unsupported context (#1382)
• feat: add Series|Expr.cum_count method (#1380)

📖 Documentation

• docs: add missing description for left join strategy (#1396)
• docs: pandas boolean content tabs (#1394)
• docs: Add page about pandas booleans (#1392)
• docs: Show ExprT as return type for Expr namespaces (#1388)
• docs: Start using Darglint (#1387)
• docs : docstrings examples - dtypes (#1121)
• docs: Document more return types (#1381)
• docs: Avoid copying >>> and ... from clipboard (#1375)

🛠️ Other improvements

• chore: update InvalidIntoExprError to mention nw.lit (#1393)
• chore: Consistent invalid IntoExpr error (#1379)
• feat: consistently raise ColumnNotFoundError for missing columns in select and drop (#1389)
• chore: suggest to add issue number in pr template (#1378)
• chore: More Polars typing, update 'used by', link to roadmap (#1377)
• chore: cleanup tpch execute (#1374)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @EdAbati, @FBruzzesi, @LiamConnors, @MarcoGorelli, @anopsy and @raisadz

Changes

:rocket: Performance improvements

• perf: simplify pandas-like with_columns (#1366)

:sparkles: Enhancements

• feat: add DataFrame.pivot for pandas like and Polars backend (#546)

🐞 Bug fixes

• fix: pandas and arrow to_dummies with nulls (#1040)

:book: Documentation

• docs: improve new_series docstring (#1365)

:hammer_and_wrench: Other improvements

• test: unxfail test_to_dummies_drop_first_na for cudf (#1364)
• chore: More typing in _polars (#1363)
• chore: Improve typing in _polars/utils (#1358)
• chore: add issue_deprecation_warning to generate_unique_token function (#1359)
• chore: refactor ArrowDataFrame.with_columns (#1345)

Thank you to all our contributors for making this release possible! @FBruzzesi, @MUKESHRAJMAHENDRAN, @MarcoGorelli and @raisadz

Changes

🚀 Performance improvements

• perf: Prefer getitem instead of loc for selecting multiple columns in pandas-like (#1355)
• perf: replace loc with getitem in Dask (#1356)

✨ Enhancements

• feat: support passing index object directly into maybe_set_index (#1319)
• feat: add more parameters to the fill_null method (#1149)
• feat: Adding support for Expr.map_batches (#1269)
• feat: pyarrow automated inference of format %Y%m%d%H%M%S (#1326)
• feat: add support for median (#1212)
• feat: allow mapping in replace_strict method (#1340)

🐞 Bug fixes

• fix: remove TODO in PandasLikeDataFrame.to_dict (#1352)
• fix: Fix nw.lit out names (#1347)

📖 Documentation

• docs: used by tabmat (#1339)

🛠️ Other improvements

• test: fix docstrings tests in CI (#1354)
• test: remove cuDF xfail for tests/series_only/to_list_test.py::test_to_list (#1353)
• chore: Rename some internals (#1351)
• ci: remove dask from tpch tests (#1348)
• ci: pin websockets, fixup 3.13 doctests, remove unnecessary xfail (#1343)
• test: Add a hypothesis test for getitem (#1098)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @DeaMariaLeon, @FBruzzesi, @IsaiasGutierrezCruz, @MarcoGorelli, @Riik, @raisadz and @sjdenny

Changes

• enabled inventory for intersphinx (#1330)

🚀 Performance improvements

• perf: improve ArrowGroupBy.__iter__ performances (#1334)

✨ Enhancements

• feat: add replace and replace_strict (#1327)
• feat: support to_list for cudf, note limitations (#1335)
• feat: add maintain_order to Expr.unique and Series.unique (#1333)

📖 Documentation

• docs: add "returns" documentation to functions.py (#1331)
• docs: update README with entry in Appears on section (#1332)

🛠️ Other improvements

• ci: trigger slow downstream libraries tests suites before release (#1321)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @EdAbati, @FBruzzesi, @MarcoGorelli, @marenwestermann, @raisadz and @srivarra

Changes

• expand "used by" (#1317)

🚀 Performance improvements

• perf: improve with_columns for pandas-like when using nw.lit (#1324)

📖 Documentation

• docs: add note about conversion as if from utc in convert_time_zone (#1320)

Thank you to all our contributors for making this release possible! @MarcoGorelli, @pre-commit-ci and @pre-commit-ci[bot]

Changes

• Revert "depr: deprecate strict in from_native / to_native in favour of pass_through" (#1311)

⚠️ Deprecations

• depr: deprecate strict in from_native / to_native in favour of pass_through (#1308)

✨ Enhancements

• feat: add is_into_dataframe (#1288)
• Enh: adding automated inferencing of format %Y-%m-%dT%H:%M in pyarrow (#1292)

🐞 Bug fixes

• fix: consistent naming of positional arguments in from_native (#1310)
• fix: narwhals_to_native_dtype raise if polars dtype is passed (#1307)
• fix: fixed-offset datetime weren't being parsed (#1303)
• fix: consistent to_numpy behaviour for tz-aware (#1305)
• fix: np.datetime64 scalar with ns resolution was become int instead of datetime in to_py_scalar (#1304)

📖 Documentation

• docs: document Stable API changes (#1309)

🛠️ Other improvements

• Revert "Revert "depr: deprecate strict in from_native / to_native in favour of pass_through (#1312)
• ci: add vegafusion in ci checks (#1228)

Thank you to all our contributors for making this release possible! @EdAbati, @FBruzzesi, @MarcoGorelli and @raisadz

Changes

✨ Enhancements

• feat: support dataframe.columns for interchange level (#1285)

🛠️ Other improvements

• ci: add marimo tests back (#1286)
• ci: set pre-commit autoupdate to run monthly instead of weekly (#1281)

Thank you to all our contributors for making this release possible! @MarcoGorelli and @raisadz

Changes

🚀 Performance improvements

• perf: make to_py_scalar 3x faster when argument is str (#1276)

✨ Enhancements

• feat: support columns and select for InterchangeFrame if _df is present (#1283)
• RFC, feat: infer datetime format for pyarrow backend (#1195)
• feat: add is_pandas_index, is_modin_index, is_cudf_index, is_pandas_like_index utility functions (#1272)
• err: replace msg = "Please set allow_series=True" with msg = "Please set allow_series=True or series_only=True" (#1280)

🐞 Bug fixes

• fix: from_native was sometimes raising unnecessarily with strict=False (#1274)

📖 Documentation

• docs: Render table of contents members ordered & removed css reference (#1275)

🛠️ Other improvements

• chore: Simplify stable api docstrings (#1278)
• ci: test with python3.13 (#1094)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @FBruzzesi, @MarcoGorelli, and @raisadz

Changes

✨ Enhancements

• feat: add .select (by str) for duckdb and ibis backend (#1266)
• feat: better error message for duplicate column names in pandas (#1270)
• feat: expose generate_temporary_column_name (#1264)

🐞 Bug fixes

• fix: remove _dtype_cache for pandas-like so that inplace modification… (#1268)
• fix: Cudf groupby iter (#1265)

Thank you to all our contributors for making this release possible! @FBruzzesi and @MarcoGorelli

Changes

🚀 Performance improvements

• perf: improve performance of Series.dtype and DataFrame.schema for pandas-like (#1255)
• perf: avoid redundant copies when resetting index for pandas (#1252)
• perf: always use copy=False when doing rename for pandas (#1247)
• perf: pandas-like fastpath in maybe_reset_index (#1246)

✨ Enhancements

• feat: Add total seconds implementation for cuDF (#1188)
• feat: add drop_null_keys argument to group_by (#1257)

📖 Documentation

• docs: add duckdb to docs (#1261)
• docs: fix parameters table rendering in quantile docs (#1251)
• docs: zen dedicated doc page (#1243)
• docs: Wimsey added to project list in Readme (#1242)
• docs: DataFrame conversion tutorial (#1240)

🛠️ Other improvements

• test: xfail cudf failures (#1259)
• test: get groupby tests passing for pandas 1.0.x (#1253)
• test: update cudf xfails based on latest version (#1250)
• test: only test dask with partitions=2 for now (#1248)
• ci: add tubular in ci checks (#1211)
• chore: make module versions as constants for use in tests (#1235)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @FBruzzesi, @LiamConnors, @MarcoGorelli, @benrutter, and @vincentarelbundock

Changes

• bug: fix pyarrow to_date (#1216)
• simplify marimo in downstream_tests.yml (#1214)
• Simplify return function in _pandas_like.utils.py (#1204)

🚀 Performance improvements

• perf: move maybe_evaluate_expr out of for loop in reuse_series_implementation (#1217)

✨ Enhancements

• feat: add is_into_series (#1215)
• enh: add dt.timestamp (#1220)
• feat: add to_py_scalar (#1194)

🐞 Bug fixes

• fix: flaky test on polars==0.20.30 (#1231)
• fix: Add df.rows(named=False) support for cuDF data frames (#1186)

📖 Documentation

• docs: Pyarrow example docstrings for Expr and stable v1 API (#1218)
• doc: add note about cuDF tests in contributing guide (#1227)
• docs: update installation.md (#1221)
• docs: move index to start of API reference (#1213)
• doc: Add uv to quick start (#1169)
• docs: Add Pyarrow example to Expr.mean (#1207)
• docs: add 'commotion' to Zen (#1208)

🛠️ Other improvements

• chore: refactor compare_dicts (#1224)
• chore: series and expr namespaces typing with generics (#1232)
• chore: nox doctests only in Python 3.12 (#1222)
• chore: use constructors (#1210)
• test: xfail iter_rows tests for cudf (#1209)
• chore: tests cleanup and add from __future__ import annotations (#1206)

Thank you to all our contributors for making this release possible! @CarloLepelaars, @DeaMariaLeon, @EdAbati, @FBruzzesi, @LiamConnors, @MarcoGorelli, @artiom-matvei, @luke396, @mscolnick, @raisadz and @schloerke

Changes

✨ Enhancements

• feat: add from_arrow (which uses the PyCapsule Interface) (#1181)
• feat: Add 'IntoSeries' (#991)
• feat: add ConstructorEager type (#1091)
• feat: Adding allow inspecting inner fields of nw.Struct (#1192)
• feat: add The Zen of Narwhals (#1168)
• feat: add <min|max>_horizontal (#1148)

🐞 Bug fixes

• fix: add __native_namespace__ for interchange support dataframes (#1172)

📖 Documentation

• docs: Update CONTRIBUTING.md with Python 3.12 (#1197)

🛠️ Other improvements

• test: update cuDF tests (#1196)
• ci: only check doctests on Python3.12 (#1189)
• ci: add downstream tests for marimo (#1166)
• chore: Delete tpch/notebooks directory (#1180)
• test: unxfail test_unary for dask (#1174)
• ci: try fixup bump-version.yml (#1171)
• ci: bump version workflow (#1063)
• chore: Use shiny's Makefile to allow for test commands to change over time (#1161)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @ChinoUkaegbu, @DeaMariaLeon, @FBruzzesi, @LiamConnors, @MarcoGorelli, @dependabot, @dependabot[bot], @lucianosrp, @luke396, @raisadz and @schloerke

Changes

✨ Enhancements

• enh: Add .rows(named=False) support for pyarrow (#1155)

🐞 Bug fixes

• fix: make Series.scatter always return new series (#1159)

📖 Documentation

• docs: Used by marimo (#1162)
• docs: separate out DataFrame and LazyFrame in api-completeness (#1157)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @Cheukting, @MarcoGorelli, @akshayka and @schloerke

Changes

🚀 Performance improvements

• perf: Impove pandas-backend indexing (#993)

✨ Enhancements

• feat: add dt.replace_time_zone (#1142)
• feat: allow format=None in str.to_datetime (#1145)
• feat: concat_str (#1128)
• feat: allow inspecting the inner type / length of nw.Array (#1136)

🐞 Bug fixes

• fix: when-then-otherwise lit string for arrow backend (#1137)
• fix: pyarrow unpivot upcast numeric (#1140)
• fix: update no_agg to work with cuDF (#1138)
• fix: fixing unnesscary raise of mean_horizontal (#1082)

📖 Documentation

• docs: apply further changes to api reference (#1150)
• docs: add missing methods in api reference (#1144)
• docs: add missing Series str methods in api reference (#1143)
• docs: fix small docs typo (#1141)

🛠️ Other improvements

• test: xfail test_unpivot_mixed_types for cuDF (#1153)

Thank you to all our contributors for making this release possible! @AlessandroMiola, @Cheukting, @DeaMariaLeon, @FBruzzesi, @LiamConnors, @MarcoGorelli, @aivanoved, @pre-commit-ci and @pre-commit-ci[bot]

Changes

✨ Enhancements

• feat: Series.str.to_datetime (#1131)
• feat: make typing available under narwhals.stable.v1 (#1130)
• feat: add nulls_last kw in dataframe sort (#1124)

🐞 Bug fixes

• fix: Fix DataFrame.__getitem__ when slicing with tuple and null slice (#1123)

📖 Documentation

• docs: add Ibis to "related projects" (#1126)
• docs: correct python bytes description (#1115)
• docs: fix some docs issues (#1114)
• docs: assorted docs updates (#1109)

🛠️ Other improvements

• ci: add py-shiny to downstream tests (#1132)
• test: xfail test_actual_object for cuDF (#1129)
• test: xfail tz_aware test for cuDF (#1118)
• test: xfail empty string test in test_unpivot (#1113)

Thank you to all our contributors for making this release possible! @FBruzzesi, @LiamConnors, @MarcoGorelli and @raisadz

Changes

• ci: add name-tests-test to pre-commit (#1093)

✨ Enhancements

• feat: allow inspecting the inner type of List (#1104)
• feat: add maybe_reset_index for pandas-like dataframe or series (#1095)
• feat: Datetime(time_unit, time_zone) and Duration(time_unit) types (#960)
• feat: add dtypes to stable api (#1087)

🐞 Bug fixes

• fix: Specialize typing in dataframe between Frame and DataFrame (#1030)
• fix: PandasLikeSeries.zip_with scalar value and DaskNamespace.when (#1099)
• fix: drop_nulls with subset of length>1 (#1090)
• fix: make narwhalify part of stable api (#1086)

📖 Documentation

• doc: add pyarrow to DataFrame docstring (#1097)
• docs: Added new logo (#1085)
• docs: add pyarrow to dataframe lazy docstring (#1055)

🛠️ Other improvements

• chore: Refactor validate_column_comparand (#1102)
• test: xfail __iter___test for cuDF (#1106)
• test: update check for cuDF series (#1105)
• test: allow cuDF series in compare_dicts (#1100)
• test: xfail test using None on to_numpy for cuDF (#1092)
• chore: rename translate_dtype to native_to_narwhals_dtype (#1089)

Thank you to all our contributors for making this release possible! @DeaMariaLeon, @FBruzzesi, @LiamConnors, @MarcoGorelli, @V-Alizade, @luke396 and @raisadz

Changes

✨ Enhancements

• feat: Initial support for nested dtypes (List, Array, Struct) (#1083)
• feat: add DataFrame|LazyFrame.unpivot method (#1043)
• feat: add to_pandas and to_arrow for interchange level of support (#1066)
• feat: add series.rename (#1073)
• feat: Adding nth() for selecting columns (#1044)

🐞 Bug fixes

• fix: fixing nw.all() doesn't work in sum_horizontal (#1075)
• fix: pandas and dask group by observed=True (#1079)
• fix: add cuDF specific implementation for join how="anti" (#1056)

🛠️ Other improvements

• chore: get_<dependency> cleanup (#1074)
• ci: Added PR labels so the Release Drafter ignores evil bots (#1069)
• ci: delete all local tags before making release (#1062)

Thank you to all our contributors for making this release possible! @Cheukting, @DeaMariaLeon, @FBruzzesi, @LiamConnors, @MarcoGorelli and @raisadz

Changes

• [pre-commit.ci] pre-commit autoupdate (#1052)
• Implementing to_native in lazyframe (#1025)
• Add pyarrow to_csv docstring (#1027)
• Extensions page improvement (#1024)
• Only run tests located in the tests directory (#1015)
• add PyArrow usage example in docstrings DataFrame.columns (#1010)

🚀 Performance improvements

• test: use pytest-randomly, recommend pytest-xdist in docs (#1019)
• perf: low-hanging overhead (#1034)

✨ Enhancements

• feat: add Series.__iter__ (#1057)
• feat: add Series.dtype return annotation (#1049)
• feat: improve Series repr (#1038)
• feat: add str.len_chars (#1036)
• feat: adding to_native to dataframe (#1014)
• feat: adding to_native to Series (#1020)

🐞 Bug fixes

• fix: move __len__ to DataFrame only (#1053)
• fix: add cuDF specific implementation for join how="anti" (#1041)
• fix: tuple supported for getitem for Pandas dataframes (#1026)
• fix: generate backend table (#1035)
• fix: polars getitem on polars==0.20.30 (#1032)
• fix: package dependency added pyarrow in docs (#1029)

📖 Documentation

• docs: add pyarrow example for dataframe getitem and to_dict (#1033)
• docs: add dtypes check in utils/check_api_reference.py (#983)
• docs: Add DataFrame.pipe to API completeness table (#1021)
• docs: Update contributing guidelines (#1031)
• docs: Include PyArrow in Tutorial/Series (#1022)
• docs: include PyArrow in Tutorial/DataFrame (#1016)
• docstrings pyarrow support for dataframe.to_pandas() (#1018)
• docs: Add pyarrow example to Installation guide (#1023)
• docs: add pycapsule interface to related projects #1011 (#1017)

🛠️ Other improvements

• fix: add cuDF specific implementation for join how="anti" (#1041)
• test: xfail test_group_by_n_unique_w_missing for cuDF (#1039)
• test: use pytest-randomly, recommend pytest-xdist in docs (#1019)
• test: add specific check for cuDF for test_to_datetime (#995)
• chore: removing docs/generate_members.py (#1012)

Thank you to all our contributors for making this release possible! @Cheukting, @Dosenpfand, @FBruzzesi, @HeitCunha, @Jupetus, @LiamConnors, @MarcoGorelli, @MathiasHolmstrom, @Morena1996, @TNieuwdorp, @V-Alizade, @anopsy, @gauthierpiarrette, @jeroenjanssens, @juanchodpg2, @lucianosrp, @mattcristal, @mikeweltevrede, @pre-commit-ci, @pre-commit-ci[bot], @sTomerG, @summerscope and @thomend

Changes

• Bug: __getitem__ check for edge cases such as subsetting 0 rows or 0 columns (#994)
• skip changelog(deps): bump astral-sh/setup-uv from 2 to 3 (#985)
• [pre-commit.ci] pre-commit autoupdate (#984)
• [pre-commit.ci] pre-commit autoupdate (#936)
• Replace assert with raise (#980)

📖 Documentation

• docs: Restructure docs a bit (#1007)
• docs: Fix link to how Narwhals works (#996)

🛠️ Other improvements

• test: convert cuDF numpy types to python types in tests (#987)
• test: add tests to check case sensitive regex for str.contains (#986)
• test: xfail case insensitive regex tests for cuDF (#982)

Thank you to all our contributors for making this release possible! @LiamConnors, @MarcoGorelli, @atulchhotray, and @raisadz

Changes

• Bug: add conversion from slice to array for selecting rows in pyarrow __getitem__ (#978)
• Debug nightly (#964)

✨ Enhancements

• feat: add DataFrame.sample method (#975)
• feat: add seed to Expr|Series.sample (#974)

📖 Documentation

• doc: Remove unnecessary paragraph in complete_example.md (#969)

🛠️ Other improvements

• test: remove xfail for cudf_constructor on str_to_uppercase test (#976)
• chore: some TODO's cleanup (#970)
• test: xfail cuDF tests that use iteration (#967)
• ci: fixup nightlies (#966)
• ci: only run nightlies on pushes to main (#965)
• ci: enable uv cache, use astral-sh official action (#956)

Thank you to all our contributors for making this release possible! @FBruzzesi, @LiamConnors, @MarcoGorelli, @aidoskanapyanov, @luke396 and @raisadz

Changes

• clean: Assorted cleanups (#958)
• Bug: fix an issue with string keys in left and semi joins in Dask, add Dask to tpch benchmarks (#957)
• bug: add support for PyArrow: multi-element __getitem__ (#949)

✨ Enhancements

• feat: Dask multiple partitions (#940)

📖 Documentation

• docs: update docstring to fix rendering in generated docs (#952)

🛠️ Other improvements

• test: replace type Any for constructor in all tests (#962)
• test: replace Any type with pytest.FixtureRequest for request inputs in all tests (#961)
• test: xfail cuDF tests for join_asof (#959)
• ci: unxfail some modin tests (#954)
• test: xfail polars>1.7.0 Expr.mode with broadcast (#953)
• test: xfail cuDF tests that use total_seconds (#951)

Thank you to all our contributors for making this release possible! @FBruzzesi, @LiamConnors, @MarcoGorelli and @raisadz

Changes

• support __getitem__ with single tuple of column names (#935)

✨ Enhancements

• feat: add Series.Scatter (#948)
• feat: add mode method (#942)
• feat: join suffix parameter (#934)

🐞 Bug fixes

• fix: group by no aggregation (#944)
• fix: Update copy param on to_numpy to default to True for cuDF (#937)

📖 Documentation

• doc: update docstrings to fix rendering in generated docs (#946)
• docs: add appears on section with links to relevant podcasts/blogs (#941)
• docs: fix docstring formatting on maybe_align_index (#938)

🛠️ Other improvements

• test: xfail to_dummies with drop_first test for cuDF (#945)
• chore: ban dask filtering with boolean mask (#939)
• test: always run tpch tests (#933)

Thank you to all our contributors for making this release possible! @FBruzzesi, @LiamConnors, @MarcoGorelli, @aidoskanapyanov and @raisadz

Changes

✨ Enhancements

• feat: implement to_arrow for cuDF (#924)
• feat: add on key to join (#927)

🐞 Bug fixes

• fix: fix join when using string literals (#930)

📖 Documentation

• doc: update drop_nulls for series documentation (#928)

🛠️ Other improvements

• feat: implement to_arrow for cuDF (#924)

Thank you to all our contributors for making this release possible! @LiamConnors, @MarcoGorelli and @raisadz