Merge branch 'main' into gpuci-query-planning

.github/workflows/additional.yml

@@ -13,7 +13,7 @@ jobs:
     timeout-minutes: 90
     steps:
       - name: Checkout source
-        uses: actions/[email protected].2
+        uses: actions/[email protected].3
 
       - name: Setup Conda Environment
         uses: conda-incubator/[email protected]
@@ -44,7 +44,7 @@ jobs:
         python-version: ["3.9", "3.10", "3.11", "3.12"]
     steps:
       - name: Checkout source
-        uses: actions/[email protected].2
+        uses: actions/[email protected].3
 
       - name: Setup Conda
         uses: conda-incubator/[email protected]

.github/workflows/conda.yml

@@ -28,7 +28,7 @@ jobs:
     name: Build (and upload)
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/[email protected].2
+      - uses: actions/[email protected].3
         with:
           fetch-depth: 0
       - name: Set up Python

.github/workflows/pre-commit.yml

@@ -11,7 +11,7 @@ jobs:
     name: pre-commit hooks
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/[email protected].2
+      - uses: actions/[email protected].3
       - uses: actions/setup-python@v5
         with:
           python-version: '3.9'

.github/workflows/release-drafter.yml

@@ -17,7 +17,7 @@ jobs:
       pull-requests: read
     runs-on: ubuntu-latest
     steps:
-      - uses: release-drafter/release-drafter@v5
+      - uses: release-drafter/release-drafter@v6
         with:
           disable-autolabeler: true
         env:

.github/workflows/release-publish.yml

@@ -20,7 +20,7 @@ jobs:
       - name: Set version env
         # Use a little bit of bash to extract the tag name from the GitHub ref
         run: echo "RELEASE_VERSION=${GITHUB_REF#refs/*/}" >> $GITHUB_ENV
-      - uses: release-drafter/release-drafter@v5
+      - uses: release-drafter/release-drafter@v6
         with:
           disable-autolabeler: true
           # Override the Release name/tag/version with the actual tag name

.github/workflows/test-report.yaml

@@ -21,7 +21,7 @@ jobs:
       run:
         shell: bash -l {0}
     steps:
-      - uses: actions/[email protected].2
+      - uses: actions/[email protected].3
         with:
           repository: dask/distributed
 
@@ -61,7 +61,7 @@ jobs:
           mv test_report.html test_short_report.html deploy/
 
       - name: Deploy 🚀
-        uses: JamesIves/github-pages-deploy-action@v4.5.0
+        uses: JamesIves/github-pages-deploy-action@v4.6.0
         with:
           branch: gh-pages
           folder: deploy

.github/workflows/tests.yml

@@ -59,7 +59,7 @@ jobs:
 
     steps:
       - name: Checkout source
-        uses: actions/[email protected].2
+        uses: actions/[email protected].3
         with:
           fetch-depth: 0  # Needed by codecov.io
 

.github/workflows/update-gpuci.yml

@@ -11,7 +11,7 @@ jobs:
     if: github.repository == 'dask/dask'
 
     steps:
-      - uses: actions/[email protected].2
+      - uses: actions/[email protected].3
 
       - name: Parse current axis YAML
         id: rapids_current

.github/workflows/upstream.yml

@@ -20,7 +20,7 @@ jobs:
     outputs:
       test-upstream: ${{ steps.detect-trigger.outputs.trigger-found }}
     steps:
-      - uses: actions/[email protected].2
+      - uses: actions/[email protected].3
         with:
           fetch-depth: 2
       - uses: xarray-contrib/ci-trigger@v1
@@ -47,7 +47,7 @@ jobs:
 
     steps:
       - name: Checkout source
-        uses: actions/[email protected].2
+        uses: actions/[email protected].3
 
       - name: Setup Conda Environment
         uses: conda-incubator/[email protected]

dask/dataframe/tests/test_categorical.py

@@ -128,6 +128,18 @@ def test_concat_unions_categoricals():
     tm.assert_frame_equal(_concat(frames5), pd.concat(frames6))
 
 
+@pytest.mark.gpu
+def test_unknown_categories_cudf():
+    # We should always start with unknown categories
+    # if `clear_known_categories` is working.
+    pytest.importorskip("dask_cudf")
+
+    with dask.config.set({"dataframe.backend": "cudf"}):
+        ddf = dd.from_dict({"a": [0, 1, 0]}, npartitions=1)
+    ddf["a"] = ddf["a"].astype("category")
+    assert not ddf["a"].cat.known
+
+
 # TODO: Remove the filterwarnings below
 @pytest.mark.parametrize(
     "numeric_only",

dask/dataframe/utils.py

@@ -23,6 +23,7 @@
 )
 from dask.dataframe._compat import PANDAS_GE_150, tm  # noqa: F401
 from dask.dataframe.dispatch import (  # noqa : F401
+    is_categorical_dtype_dispatch,
     make_meta,
     make_meta_obj,
     meta_nonempty,
@@ -283,22 +284,22 @@ def clear_known_categories(x, cols=None, index=True, dtype_backend=None):
         # categorical accessor is not yet available
         return x
 
-    if isinstance(x, (pd.Series, pd.DataFrame)):
+    if not is_index_like(x):
         x = x.copy()
-        if isinstance(x, pd.DataFrame):
+        if is_dataframe_like(x):
             mask = x.dtypes == "category"
             if cols is None:
                 cols = mask[mask].index
             elif not mask.loc[cols].all():
                 raise ValueError("Not all columns are categoricals")
             for c in cols:
                 x[c] = x[c].cat.set_categories([UNKNOWN_CATEGORIES])
-        elif isinstance(x, pd.Series):
-            if isinstance(x.dtype, pd.CategoricalDtype):
+        elif is_series_like(x):
+            if is_categorical_dtype_dispatch(x.dtype):
                 x = x.cat.set_categories([UNKNOWN_CATEGORIES])
-        if index and isinstance(x.index, pd.CategoricalIndex):
+        if index and is_categorical_dtype_dispatch(x.index.dtype):
             x.index = x.index.set_categories([UNKNOWN_CATEGORIES])
-    elif isinstance(x, pd.CategoricalIndex):
+    elif is_categorical_dtype_dispatch(x.dtype):
         x = x.set_categories([UNKNOWN_CATEGORIES])
     return x