Tamas Mate has posted comments on this change. ( http://gerrit.cloudera.org:8080/15835 )
Change subject: IMPALA-9699: Skip '-1' values when aggregating num_null statistics ...................................................................... Patch Set 3: (2 comments) Hi Csaba, Sahil, Thank you for the quick review, I have answered your comments. Let me know your thoughts. http://gerrit.cloudera.org:8080/#/c/15835/3/be/src/exec/incr-stats-util.cc File be/src/exec/incr-stats-util.cc: http://gerrit.cloudera.org:8080/#/c/15835/3/be/src/exec/incr-stats-util.cc@128 PS3, Line 128: if (num_new_nulls >= 0) num_nulls += num_new_nulls; > agree with Csaba I will try to address this comment, but it is possible that I misunderstood the concerns. This 'Update' method is only called when 'num_nulls' are available for at least one partition in 'FinalizePartitionedColumnStats', in which the newly counted and existing partition stats are used to aggregate the partition stats to column statistics. I think the final 'num_nulls' result of the consecutive 'Update' calls can only be '>=0' value and it should indicate the number of null values for the partitions where incremental stats are available. http://gerrit.cloudera.org:8080/#/c/15835/3/be/src/exec/incr-stats-util.cc@130 PS3, Line 130: avg_width += (new_avg_width * num_new_rows); > This is not new code, but it looks very wrong - there is a member called to It looks like we divide the avg_width by num_rows in line 137. I would need some more time to understand the reason behind this logic and the total_width, is it ok if I open a new Jira to untangle this part? -- To view, visit http://gerrit.cloudera.org:8080/15835 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I3fdf48a6c88378145078e068e12ade48c460f956 Gerrit-Change-Number: 15835 Gerrit-PatchSet: 3 Gerrit-Owner: Tamas Mate <tm...@cloudera.com> Gerrit-Reviewer: Csaba Ringhofer <csringho...@cloudera.com> Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com> Gerrit-Reviewer: Sahil Takiar <stak...@cloudera.com> Gerrit-Reviewer: Tamas Mate <tm...@cloudera.com> Gerrit-Comment-Date: Thu, 30 Apr 2020 09:38:09 +0000 Gerrit-HasComments: Yes
