try filter fastpath

narwhals/pandas_like/dataframe.py

@@ -94,16 +94,23 @@ def select(
 
     def filter(
         self,
-        *predicates: IntoPandasExpr | Iterable[IntoPandasExpr],
+        *predicates: Iterable[IntoPandasExpr],
     ) -> Self:
-        from narwhals.pandas_like.namespace import PandasNamespace
-
-        plx = PandasNamespace(self._implementation)
-        expr = plx.all_horizontal(*predicates)
-        # Safety: all_horizontal's expression only returns a single column.
-        mask = expr._call(self)[0]
-        _mask = validate_dataframe_comparand(mask)
-        return self._from_dataframe(self._dataframe[_mask])
+        masks = evaluate_into_exprs(self, *predicates)
+        if len(masks) == 1:
+            return self._from_dataframe(
+                self._dataframe[validate_dataframe_comparand(masks[0])]
+            )
+
+        return self._from_dataframe(
+            self._dataframe[
+                validate_dataframe_comparand(
+                    horizontal_concat(masks, implementation=self._implementation).all(
+                        axis=1
+                    )
+                )
+            ]
+        )
 
     def with_columns(
         self,