Set pyarrow string conversion

Author: milesgranger

tests/conftest.py

@@ -691,6 +691,12 @@ def configure_shuffling(shuffle_method):
         yield
 
 
+@pytest.fixture
+def configure_use_pyarrow_strings():
+    with dask.config.set({"dataframe.convert-string": True}):
+        yield
+
+
 # Include https://github.com/dask/distributed/pull/7534
 P2P_RECHUNK_AVAILABLE = Version(distributed.__version__) >= Version("2023.2.1")
 

tests/workflows/test_from_csv_to_parquet.py

@@ -102,7 +102,9 @@ def from_csv_to_parquet_client(
 )
 
 
-def test_from_csv_to_parquet(from_csv_to_parquet_client, s3_factory, s3_url):
+def test_from_csv_to_parquet(
+    from_csv_to_parquet_client, s3_factory, s3_url, configure_use_pyarrow_strings
+):
     s3 = s3_factory(anon=True)
     files = s3.ls("s3://gdelt-open-data/events/")[:1000]
     files = [f"s3://{f}" for f in files]