Follow up making Python tests more deterministic

.pre-commit-config.yaml

@@ -86,12 +86,8 @@ repos:
         types: [python]
       - id: no-unseeded-default-rng
         name: no-unseeded-default-rng
-        description: 'Enforce that no non-seeded default_rng is used and default_rng is used instead of np.random.seed'
-        entry: |
-          # Check for usage of default_rng without seeding
-          default_rng\(\)|
-          # Check for usage of np.random.seed
-          np.random.seed\(
+        description: 'Enforce that np.random.default_rng is seeded'
+        entry: default_rng\(\)
         language: pygrep
         types: [python]
       - id: cmake-format

python/cudf/cudf/tests/test_parquet.py

@@ -193,11 +193,6 @@ def parquet_file(request, tmp_path_factory, pdf):
     return fname
 
 
-@pytest.fixture(scope="module")
-def rdg_seed():
-    return int(os.environ.get("TEST_CUDF_RDG_SEED", "42"))
-
-
 def make_pdf(nrows, ncolumns=1, nvalids=0, dtype=np.int64):
     test_pdf = pd.DataFrame(
         [list(range(ncolumns * i, ncolumns * (i + 1))) for i in range(nrows)],
@@ -431,7 +426,7 @@ def num_row_groups(rows, group_size):
         assert a == b
 
 
-def test_parquet_read_filtered(tmpdir, rdg_seed):
+def test_parquet_read_filtered(tmpdir):
     # Generate data
     fname = tmpdir.join("filtered.parquet")
     dg.generate(
@@ -455,13 +450,13 @@ def test_parquet_read_filtered(tmpdir, rdg_seed):
                 dg.ColumnParameters(
                     40,
                     0.2,
-                    lambda: np.random.default_rng(seed=None).integers(
+                    lambda: np.random.default_rng(seed=0).integers(
                         0, 100, size=40
                     ),
                     True,
                 ),
             ],
-            seed=rdg_seed,
+            seed=42,
         ),
         format={"name": "parquet", "row_group_size": 64},
     )

python/dask_cudf/dask_cudf/tests/test_reductions.py

@@ -1,7 +1,5 @@
 # Copyright (c) 2021-2024, NVIDIA CORPORATION.
 
-import numpy as np
-import pandas as pd
 import pytest
 
 import dask
@@ -10,20 +8,7 @@
 import cudf
 
 import dask_cudf
-
-
-def _make_random_frame(nelem, npartitions=2):
-    rng = np.random.default_rng(seed=0)
-    df = pd.DataFrame(
-        {
-            "x": rng.integers(0, 5, size=nelem),
-            "y": rng.normal(loc=1.0, scale=1.0, size=nelem),
-        }
-    )
-    gdf = cudf.DataFrame.from_pandas(df)
-    dgf = dask_cudf.from_cudf(gdf, npartitions=npartitions)
-    return df, dgf
-
+from dask_cudf.tests.utils import _make_random_frame
 
 _reducers = ["sum", "count", "mean", "var", "std", "min", "max"]
 

python/dask_cudf/dask_cudf/tests/utils.py

@@ -19,7 +19,7 @@
 
 
 def _make_random_frame(nelem, npartitions=2, include_na=False):
-    rng = np.random.default_rng(seed=None)
+    rng = np.random.default_rng(seed=0)
     df = pd.DataFrame(
         {"x": rng.random(size=nelem), "y": rng.random(size=nelem)}
     )