Github user gatorsmile commented on a diff in the pull request:
https://github.com/apache/spark/pull/19083#discussion_r142020435
--- Diff: sql/core/src/test/resources/sql-tests/inputs/group-by.sql ---
@@ -30,8 +30,15 @@ SELECT a + 2, COUNT(b) FROM testData GROUP BY a + 1;
SELECT a + 1 + 1, COUNT(b) FROM testData GROUP BY a + 1;
-- Aggregate with nulls.
+--
+-- In SPARK-21871, we added code to check the actual bytecode size of
gen'd methods. If the size
+-- goes over `hugeMethodLimit`, Spark fails to compile the methods and the
execution also fails
+-- in a test mode. So, we explicitly turn off whole-stage codegen here.
+-- This guard can be removed if this issue fixed.
+SET spark.sql.codegen.wholeStage=false;
--- End diff --
Please change the value of `spark.sql.codegen.hugeMethodLimit`
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
