Tim Armstrong has posted comments on this change. Change subject: IMPALA-5363: Reset probe_batch_ after reaching limit. ......................................................................
Patch Set 1: (2 comments) http://gerrit.cloudera.org:8080/#/c/7014/1//COMMIT_MSG Commit Message: I think this patch needs a test case (I think you already had a query, right). We should also check to see if the same bug applies to nested loop join node. The same query with the equality join condition removed might work to produce the same plan with a nested loop join. http://gerrit.cloudera.org:8080/#/c/7014/1/be/src/exec/partitioned-hash-join-node.cc File be/src/exec/partitioned-hash-join-node.cc: Line 506: while (true) { How about we avoid the duplicate ReachedLimit() return path by changing this loop condition to "while (!ReachedLimit())" and handling both cases in the same place? This would mean we evaluate ReachedLimit() more times but each loop iteration processes many rows so I doubt the overhead is significant. -- To view, visit http://gerrit.cloudera.org:8080/7014 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: Iafd621d33a4e2fac42391504566ffd8dd0e18a67 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: anujphadke <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-Reviewer: anujphadke <[email protected]> Gerrit-HasComments: Yes
