comments

Author: TrevorBergeron

bigframes/session/direct_gbq_execution.py

@@ -25,8 +25,9 @@
 
 
 # used only in testing right now, BigQueryCachingExecutor is the fully featured engine
-# simplified, doest not do large >10 gb result queries, error handling, respect global config
-# or record metrics
+# simplified, doesnt not do large >10 gb result queries, error handling, respect global config
+# or record metrics. Also avoids caching, and most pre-compile rewrites, to better serve as a
+# reference for validating more complex executors.
 class DirectGbqExecutor(semi_executor.SemiExecutor):
     def __init__(self, bqclient: bigquery.Client):
         self.bqclient = bqclient

tests/system/small/engines/test_read_local.py

@@ -20,6 +20,7 @@
 
 pytest.importorskip("polars")
 
+# Polars used as reference as its fast and local. Generally though, prefer gbq engine where they disagree.
 REFERENCE_ENGINE = polars_executor.PolarsExecutor()