[
https://issues.apache.org/jira/browse/HIVE-5276?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13816875#comment-13816875
]
Navis commented on HIVE-5276:
-----------------------------
[[~brocknoland] Ah, sorry. When I've been doing this, I've assumed that one
input row would result one to zero output row, which is generally right. But
with second thought, it's possible fetch task to have UDTF Operator which would
make arbitrary number of output rows. So I've changed the patch status. I'll
update this later.
> Skip useless string encoding stage for hiveserver2
> --------------------------------------------------
>
> Key: HIVE-5276
> URL: https://issues.apache.org/jira/browse/HIVE-5276
> Project: Hive
> Issue Type: Improvement
> Components: HiveServer2
> Reporter: Navis
> Assignee: Navis
> Priority: Minor
> Attachments: HIVE-5276.3.patch.txt, HIVE-5276.4.patch.txt
>
>
> Current hiveserver2 acquires rows in string format which is used for cli
> output. Then convert them into row again and convert to final format lastly.
> This is inefficient and memory consuming.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
