ngsg commented on code in PR #4190:
URL: https://github.com/apache/hive/pull/4190#discussion_r1170991803


##########
ql/src/test/queries/clientpositive/antijoin2.q:
##########
@@ -0,0 +1,75 @@
+set hive.merge.nway.joins=false;
+set hive.vectorized.execution.enabled=false;
+set hive.auto.convert.join=true;
+set hive.auto.convert.anti.join=true;
+
+drop table if exists tt1;
+drop table if exists tt2;
+drop table if exists tt3;
+
+create table tt1 (ws_order_number bigint, ws_ext_ship_cost decimal(7, 2));
+create table tt2 (ws_order_number bigint);
+create table tt3 (wr_order_number bigint);
+
+insert into tt1 values (42, 3093.96), (1041, 299.28), (1378, 85.56), (1378, 
719.44), (1395, 145.68);
+insert into tt2 values (1378), (1395);
+insert into tt3 values (42), (1041);
+
+-- The result should be the same regardless of vectorization.
+
+explain

Review Comment:
   I updated the qfile and expected output for verifying MergeJoin -> MapJoin 
pattern.
   
   However, I think this issue is unrelated with MergeJoin -> MapJoin pattern 
and depends only on MapJoinOperator. For example, the second query in this 
qfile contains only one Join, but the current Hive returns no rows although tt1 
- tt2 is not empty. Also, I have checked that the current Hive returns wrong 
result for the first query without MergeJoin -> MapJoin pattern. (I used mvn 
test to run both of the queries.)
   
   I think you could reproduce this issue with MapJoin -> MapJoin pattern if 
you disable vectorized execution. Could you please give me more details about 
the configuration that you used for reproducing the problem? Also I would be 
glad if you could run this qfile in your local environment and share the result.
   



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to