more tests, prettier sql

bigframes/core/compile/compiled.py

@@ -1277,11 +1277,11 @@ def to_sql(
         )
         if sorted:
             sql = textwrap.dedent(
-                f"""
-                SELECT * EXCEPT (`{offsets_id}`)
-                FROM ({sql})
-                ORDER BY `{offsets_id}`
-                """
+                f"SELECT * EXCEPT (`{offsets_id}`)\n"
+                "FROM (\n"
+                f"{sql}\n"
+                ")\n"
+                f"ORDER BY `{offsets_id}`\n"
             )
         return typing.cast(str, sql)
 

bigframes/series.py

@@ -287,8 +287,8 @@ def to_pandas(
                 take longer to execute and require more computation. If set to a value other than
                 None, this will supersede the global config.
             ordered (bool, default True):
-                Determines whether the resulting pandas series will be ordered. In some cases,
-                unordered may result in a faster-executing query.
+                Determines whether the resulting pandas series will be deterministically ordered.
+                In some cases, unordered may result in a faster-executing query.
 
 
         Returns:

tests/system/large/ml/test_cluster.py

@@ -105,7 +105,7 @@ def test_cluster_configure_fit_score_predict(
         index=pd.Index(["test1", "test2", "test3", "test4"], dtype="string[pyarrow]"),
     )
     expected.index.name = "observation"
-    assert_pandas_df_equal(result, expected)
+    assert_pandas_df_equal(result, expected, ignore_order=True)
 
     # save, load, check n_clusters to ensure configuration was kept
     reloaded_model = model.to_gbq(

tests/system/small/test_groupby.py

@@ -16,6 +16,7 @@
 import pytest
 
 import bigframes.pandas as bpd
+from tests.system.utils import assert_pandas_df_equal
 
 
 @pytest.mark.parametrize(
@@ -88,16 +89,23 @@ def test_dataframe_groupby_aggregate(
     pd.testing.assert_frame_equal(pd_result, bf_result_computed, check_dtype=False)
 
 
-def test_dataframe_groupby_agg_string(scalars_df_index, scalars_pandas_df_index):
+@pytest.mark.parametrize(
+    ("ordered"),
+    [
+        (True),
+        (False),
+    ],
+)
+def test_dataframe_groupby_agg_string(
+    scalars_df_index, scalars_pandas_df_index, ordered
+):
     col_names = ["int64_too", "float64_col", "int64_col", "bool_col", "string_col"]
     bf_result = scalars_df_index[col_names].groupby("string_col").agg("count")
     pd_result = scalars_pandas_df_index[col_names].groupby("string_col").agg("count")
-    bf_result_computed = bf_result.to_pandas()
+    bf_result_computed = bf_result.to_pandas(ordered=ordered)
 
-    pd.testing.assert_frame_equal(
-        pd_result,
-        bf_result_computed,
-        check_dtype=False,
+    assert_pandas_df_equal(
+        pd_result, bf_result_computed, check_dtype=False, ignore_order=not ordered
     )
 
 
@@ -270,13 +278,22 @@ def test_dataframe_groupby_kurt(scalars_df_index, scalars_pandas_df_index):
     pd.testing.assert_frame_equal(pd_result, bf_result, check_dtype=False)
 
 
-def test_dataframe_groupby_diff(scalars_df_index, scalars_pandas_df_index):
+@pytest.mark.parametrize(
+    ("ordered"),
+    [
+        (True),
+        (False),
+    ],
+)
+def test_dataframe_groupby_diff(scalars_df_index, scalars_pandas_df_index, ordered):
     col_names = ["float64_col", "int64_col", "string_col"]
     bf_result = scalars_df_index[col_names].groupby("string_col").diff(-1)
     pd_result = scalars_pandas_df_index[col_names].groupby("string_col").diff(-1)
-    bf_result_computed = bf_result.to_pandas()
+    bf_result_computed = bf_result.to_pandas(ordered=ordered)
 
-    pd.testing.assert_frame_equal(pd_result, bf_result_computed, check_dtype=False)
+    assert_pandas_df_equal(
+        pd_result, bf_result_computed, check_dtype=False, ignore_order=not ordered
+    )
 
 
 def test_dataframe_groupby_getitem(