test: use skip tests on old versions instead of xfailing, print slowest tests (#2375)

MarcoGorelli · web-flow · commit d63ec65ef400 · 2025-04-12T11:43:35.000+01:00
diff --git a/.github/workflows/pytest.yml b/.github/workflows/pytest.yml
@@ -58,7 +58,7 @@ jobs:
         run: uv pip freeze
       - name: Run pytest
         run: |
-          pytest tests --cov=narwhals --cov=tests --runslow --cov-fail-under=95 --constructors=pandas,pandas[nullable],pandas[pyarrow],pyarrow,modin[pyarrow],polars[eager],polars[lazy],dask,duckdb,sqlframe
+          pytest tests --cov=narwhals --cov=tests --runslow --cov-fail-under=95 --constructors=pandas,pandas[nullable],pandas[pyarrow],pyarrow,modin[pyarrow],polars[eager],polars[lazy],dask,duckdb,sqlframe --durations=30
 
   pytest-full-coverage:
     strategy:
@@ -94,7 +94,7 @@ jobs:
       - name: show-deps
         run: uv pip freeze
       - name: Run pytest
-        run: pytest tests --cov=narwhals --cov=tests --cov-fail-under=100 --runslow --constructors=pandas,pandas[nullable],pandas[pyarrow],pyarrow,modin[pyarrow],polars[eager],polars[lazy],dask,duckdb,sqlframe
+        run: pytest tests --cov=narwhals --cov=tests --cov-fail-under=100 --runslow --constructors=pandas,pandas[nullable],pandas[pyarrow],pyarrow,modin[pyarrow],polars[eager],polars[lazy],dask,duckdb,sqlframe --durations=30
       - name: Run doctests
         # reprs differ between versions, so we only run doctests on the latest Python
         if: matrix.python-version == '3.13'
diff --git a/.github/workflows/typing.yml b/.github/workflows/typing.yml
@@ -9,7 +9,7 @@ env:
   PY_COLORS: 1
 
 jobs:
-  mypy:
+  typing:
     strategy:
       matrix:
         python-version: ["3.12"]
diff --git a/narwhals/_arrow/series.py b/narwhals/_arrow/series.py
@@ -1087,8 +1087,8 @@ def _hist_from_bins(bins: Sequence[int | float]):  # type: ignore[no-untyped-def
             bin_right = bins[1:]
             return counts, bin_right
 
-        counts: Sequence[int | float] | np.typing.ArrayLike
-        bin_right: Sequence[int | float] | np.typing.ArrayLike
+        counts: Sequence[int | float | pa.Scalar[Any]] | np.typing.ArrayLike
+        bin_right: Sequence[int | float | pa.Scalar[Any]] | np.typing.ArrayLike
 
         data_count = pc.sum(
             pc.invert(pc.or_(pc.is_nan(self.native), pc.is_null(self.native))).cast(
diff --git a/tests/expr_and_series/all_horizontal_test.py b/tests/expr_and_series/all_horizontal_test.py
@@ -51,12 +51,9 @@ def test_allh_all(constructor: Constructor) -> None:
     assert_equal_data(result, expected)
 
 
-def test_allh_nth(
-    constructor: Constructor,
-    request: pytest.FixtureRequest,
-) -> None:
+def test_allh_nth(constructor: Constructor) -> None:
     if "polars" in str(constructor) and POLARS_VERSION < (1, 0):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     data = {
         "a": [False, False, True],
         "b": [False, True, True],
diff --git a/tests/expr_and_series/arithmetic_test.py b/tests/expr_and_series/arithmetic_test.py
@@ -69,7 +69,7 @@ def test_right_arithmetic_expr(
     request: pytest.FixtureRequest,
 ) -> None:
     if "dask" in str(constructor) and DASK_VERSION < (2024, 10):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     if attr == "__rmod__" and any(
         x in str(constructor) for x in ["pandas_pyarrow", "modin_pyarrow"]
     ):
diff --git a/tests/expr_and_series/binary_test.py b/tests/expr_and_series/binary_test.py
@@ -8,9 +8,9 @@
 from tests.utils import assert_equal_data
 
 
-def test_expr_binary(constructor: Constructor, request: pytest.FixtureRequest) -> None:
+def test_expr_binary(constructor: Constructor) -> None:
     if "dask" in str(constructor) and DASK_VERSION < (2024, 10):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     data = {"a": [1, 3, 2], "b": [4, 4, 6], "z": [7.0, 8.0, 9.0]}
     df_raw = constructor(data)
     result = nw.from_native(df_raw).with_columns(
diff --git a/tests/expr_and_series/cast_test.py b/tests/expr_and_series/cast_test.py
@@ -272,7 +272,7 @@ def test_cast_struct(request: pytest.FixtureRequest, constructor: Constructor) -
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     data = {
         "a": [
@@ -322,7 +322,7 @@ def test_raise_if_polars_dtype(constructor: Constructor) -> None:
 
 def test_cast_time(request: pytest.FixtureRequest, constructor: Constructor) -> None:
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     if any(
         backend in str(constructor) for backend in ("dask", "pyspark", "modin", "cudf")
@@ -337,7 +337,7 @@ def test_cast_time(request: pytest.FixtureRequest, constructor: Constructor) ->
 
 def test_cast_binary(request: pytest.FixtureRequest, constructor: Constructor) -> None:
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     if any(backend in str(constructor) for backend in ("cudf", "dask", "modin")):
         request.applymarker(pytest.mark.xfail)
diff --git a/tests/expr_and_series/cat/get_categories_test.py b/tests/expr_and_series/cat/get_categories_test.py
@@ -11,12 +11,9 @@
 data = {"a": ["one", "two", "two"]}
 
 
-def test_get_categories(
-    request: pytest.FixtureRequest,
-    constructor_eager: ConstructorEager,
-) -> None:
+def test_get_categories(constructor_eager: ConstructorEager) -> None:
     if "pyarrow_table" in str(constructor_eager) and PYARROW_VERSION < (15, 0, 0):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     df = nw.from_native(constructor_eager(data), eager_only=True)
     df = df.select(nw.col("a").cast(nw.Categorical))
diff --git a/tests/expr_and_series/list/len_test.py b/tests/expr_and_series/list/len_test.py
@@ -21,7 +21,7 @@ def test_len_expr(
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     result = nw.from_native(constructor(data)).select(
         nw.col("a").cast(nw.List(nw.Int32())).list.len()
@@ -38,7 +38,7 @@ def test_len_series(
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor_eager) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     df = nw.from_native(constructor_eager(data), eager_only=True)
 
diff --git a/tests/expr_and_series/nth_test.py b/tests/expr_and_series/nth_test.py
@@ -25,10 +25,9 @@ def test_nth(
     constructor: Constructor,
     idx: int | list[int],
     expected: dict[str, list[int]],
-    request: pytest.FixtureRequest,
 ) -> None:
     if "polars" in str(constructor) and POLARS_VERSION < (1, 0, 0):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     df = nw.from_native(constructor(data))
     result = df.select(nw.nth(idx))
     assert_equal_data(result, expected)
diff --git a/tests/expr_and_series/rolling_std_test.py b/tests/expr_and_series/rolling_std_test.py
@@ -71,7 +71,6 @@
 )
 @pytest.mark.parametrize("kwargs_and_expected", kwargs_and_expected)
 def test_rolling_std_expr(
-    request: pytest.FixtureRequest,
     constructor_eager: ConstructorEager,
     kwargs_and_expected: dict[str, Any],
 ) -> None:
@@ -80,7 +79,7 @@ def test_rolling_std_expr(
     expected = kwargs_and_expected["expected"]
 
     if "polars" in str(constructor_eager) and POLARS_VERSION < (1,):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     df = nw.from_native(constructor_eager(data))
     result = df.select(nw.col("a").rolling_std(**kwargs).alias(name))
@@ -93,12 +92,11 @@ def test_rolling_std_expr(
 )
 @pytest.mark.parametrize("kwargs_and_expected", kwargs_and_expected)
 def test_rolling_std_series(
-    request: pytest.FixtureRequest,
     constructor_eager: ConstructorEager,
     kwargs_and_expected: dict[str, Any],
 ) -> None:
     if "polars" in str(constructor_eager) and POLARS_VERSION < (1,):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     name = kwargs_and_expected["name"]
     kwargs = kwargs_and_expected["kwargs"]
diff --git a/tests/expr_and_series/rolling_var_test.py b/tests/expr_and_series/rolling_var_test.py
@@ -88,12 +88,11 @@ def test_rolling_var_expr(
 )
 @pytest.mark.parametrize("kwargs_and_expected", kwargs_and_expected)
 def test_rolling_var_series(
-    request: pytest.FixtureRequest,
     constructor_eager: ConstructorEager,
     kwargs_and_expected: dict[str, Any],
 ) -> None:
     if "polars" in str(constructor_eager) and POLARS_VERSION < (1,):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     name = kwargs_and_expected["name"]
     kwargs = kwargs_and_expected["kwargs"]
diff --git a/tests/frame/explode_test.py b/tests/frame/explode_test.py
@@ -43,7 +43,7 @@ def test_explode_single_col(
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     result = (
         nw.from_native(constructor(data))
@@ -93,7 +93,7 @@ def test_explode_multiple_cols(
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     result = (
         nw.from_native(constructor(data))
@@ -114,7 +114,7 @@ def test_explode_shape_error(
         request.applymarker(pytest.mark.xfail)
 
     if "pandas" in str(constructor) and PANDAS_VERSION < (2, 2):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     with pytest.raises(
         (ShapeError, NotImplementedError),
@@ -136,7 +136,7 @@ def test_explode_invalid_operation_error(
         request.applymarker(pytest.mark.xfail)
 
     if "polars" in str(constructor) and POLARS_VERSION < (0, 20, 6):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
 
     with pytest.raises(
         InvalidOperationError,
diff --git a/tests/frame/join_test.py b/tests/frame/join_test.py
@@ -180,9 +180,9 @@ def test_inner_join_single_key(constructor: Constructor) -> None:
     assert_equal_data(result_on, expected)
 
 
-def test_cross_join(constructor: Constructor, request: pytest.FixtureRequest) -> None:
+def test_cross_join(constructor: Constructor) -> None:
     if "duckdb" in str(constructor) and DUCKDB_VERSION < (1, 1, 4):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     data = {"antananarivo": [1, 3, 2]}
     df = nw.from_native(constructor(data))
     result = df.join(df, how="cross").sort("antananarivo", "antananarivo_right")  # type: ignore[arg-type]
@@ -220,11 +220,9 @@ def test_suffix(constructor: Constructor, how: str, suffix: str) -> None:
 
 
 @pytest.mark.parametrize("suffix", ["_right", "_custom_suffix"])
-def test_cross_join_suffix(
-    constructor: Constructor, suffix: str, request: pytest.FixtureRequest
-) -> None:
+def test_cross_join_suffix(constructor: Constructor, suffix: str) -> None:
     if "duckdb" in str(constructor) and DUCKDB_VERSION < (1, 1, 4):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     data = {"antananarivo": [1, 3, 2]}
     df = nw.from_native(constructor(data))
     result = df.join(df, how="cross", suffix=suffix).sort(  # type: ignore[arg-type]
diff --git a/tests/frame/select_test.py b/tests/frame/select_test.py
@@ -117,13 +117,11 @@ def test_missing_columns(
             df.select(nw.col("fdfa"))
 
 
-def test_left_to_right_broadcasting(
-    constructor: Constructor, request: pytest.FixtureRequest
-) -> None:
+def test_left_to_right_broadcasting(constructor: Constructor) -> None:
     if "duckdb" in str(constructor) and DUCKDB_VERSION < (1, 3):
         pytest.skip()
     if "dask" in str(constructor) and DASK_VERSION < (2024, 10):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     df = nw.from_native(constructor({"a": [1, 1, 2], "b": [4, 5, 6]}))
     result = df.select(nw.col("a") + nw.col("b").sum())
     expected = {"a": [16, 16, 17]}
diff --git a/tests/frame/with_columns_test.py b/tests/frame/with_columns_test.py
@@ -62,7 +62,7 @@ def test_with_columns_dtypes_single_row(
     request: pytest.FixtureRequest,
 ) -> None:
     if "pyarrow_table" in str(constructor) and PYARROW_VERSION < (15,):
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     if ("pyspark" in str(constructor)) or "duckdb" in str(constructor):
         request.applymarker(pytest.mark.xfail)
     data = {"a": ["foo"]}
diff --git a/tests/hypothesis/basic_arithmetic_test.py b/tests/hypothesis/basic_arithmetic_test.py
diff --git a/tests/selectors_test.py b/tests/selectors_test.py
@@ -96,6 +96,8 @@ def test_datetime(constructor: Constructor, request: pytest.FixtureRequest) -> N
         or ("pandas" in str(constructor) and PANDAS_VERSION < (2,))
     ):
         request.applymarker(pytest.mark.xfail)
+    if "modin" in str(constructor):
+        pytest.skip(reason="too slow")
 
     ts1 = datetime(2000, 11, 20, 18, 12, 16, 600000)
     ts2 = datetime(2020, 10, 30, 10, 20, 25, 123000)
@@ -228,15 +230,12 @@ def test_set_ops(
     assert sorted(result) == expected
 
 
-def test_subtract_expr(
-    constructor: Constructor,
-    request: pytest.FixtureRequest,
-) -> None:
+def test_subtract_expr(constructor: Constructor) -> None:
     if "polars" in str(constructor) and POLARS_VERSION < (0, 20, 27):
         # In old Polars versions, cs.numeric() - col('a')
         # would exclude column 'a' from the result, as opposed to
         # subtracting it.
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     df = nw.from_native(constructor(data))
     result = df.select(ncs.numeric() - nw.col("a"))
     expected = {"a": [0, 0, 0], "c": [3.1, 4.0, 4.0]}
@@ -263,10 +262,10 @@ def test_set_ops_invalid(constructor: Constructor) -> None:
 def test_tz_aware(constructor: Constructor, request: pytest.FixtureRequest) -> None:
     if "polars" in str(constructor) and POLARS_VERSION < (1, 19):
         # bug in old polars
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     if "pyarrow_table" in str(constructor) and PYARROW_VERSION < (12,):
         # bug in old pyarrow
-        request.applymarker(pytest.mark.xfail)
+        pytest.skip()
     if "duckdb" in str(constructor) or "pyspark" in str(constructor):
         # replace_time_zone not implemented
         request.applymarker(pytest.mark.xfail)
diff --git a/tests/series_only/hist_test.py b/tests/series_only/hist_test.py
diff --git a/tests/tpch_q1_test.py b/tests/tpch_q1_test.py
