Add nans_to_nulls to Frame

python/cudf/cudf/core/frame.py

@@ -181,6 +181,18 @@ def deserialize(cls, header, frames):
         )
         return cls._from_data(col_accessor)
 
+    @_performance_tracking
+    def nans_to_nulls(self):
+        result = []
+        for col in self._columns:
+            converted = col.nans_to_nulls()
+            if converted is col:
+                converted = converted.copy()
+            result.append(converted)
+        return self._from_data_like_self(
+            self._data._from_columns_like_self(result)
+        )
+
     @classmethod
     @_performance_tracking
     def _from_data(cls, data: MutableMapping) -> Self:

python/cudf/cudf/core/indexed_frame.py

@@ -1935,15 +1935,7 @@ def nans_to_nulls(self):
         1  <NA>  3.14
         2  <NA>  <NA>
         """
-        result = []
-        for col in self._columns:
-            converted = col.nans_to_nulls()
-            if converted is col:
-                converted = converted.copy()
-            result.append(converted)
-        return self._from_data_like_self(
-            self._data._from_columns_like_self(result)
-        )
+        return super().nans_to_nulls()
 
     @_performance_tracking
     def interpolate(

python/cudf/cudf/tests/dataframe/methods/test_sort_values.py

@@ -228,3 +228,17 @@ def test_dataframe_loc_duplicate_index_scalar():
     gdf_sorted = gdf.sort_values(by=list(gdf.columns), axis=0)
 
     assert_eq(pdf_sorted, gdf_sorted)
+
+
+def test_dataframe_sort_values_multindex():
+    df = cudf.DataFrame(
+        {"a": [2, 1, 2, 1], "b": [1, 2, 1, 2], "c": [4, 3, 2, 1]}
+    )
+    df = df.set_index(["a", "b"])
+    pdf = df.to_pandas()
+
+    expected = pdf.sort_values(by="a", ascending=True)
+    with cudf.option_context("mode.pandas_compatible", True):
+        result = df.sort_values(by="a", ascending=True)
+
+    assert_eq(expected, result)