[
https://issues.apache.org/jira/browse/HIVE-27115?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17697152#comment-17697152
]
Yi Zhang commented on HIVE-27115:
---------------------------------
[~rajesh.balamohan] Is there scenario where in Tez mode multiple
TableScanOperators passed to HiveInputFormat? I assume in Tez each TS has its
own input initialized so it doesn't run into this issue. Understand that MR is
deprecated, however the logic of HiveInputFormat itself has this bug when
combines multiple TS.
In some use cases of storagehandlers that was added in MR mode have some issues
when run in Tez mode.
> HiveInputFormat column project push down wrong fields (MR)
> ----------------------------------------------------------
>
> Key: HIVE-27115
> URL: https://issues.apache.org/jira/browse/HIVE-27115
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 3.1.3, 4.0.0-alpha-2
> Reporter: Yi Zhang
> Assignee: Yi Zhang
> Priority: Major
> Labels: pull-request-available
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> For query such as
> select * from (
> select r_name from r
> union all
> select t_name from t
> ) unioned
>
> in MR execution, when column project push down for splits, t_name gets pushed
> down to table r.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
