[
https://issues.apache.org/jira/browse/HIVE-11281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14633945#comment-14633945
]
Gunther Hagleitner commented on HIVE-11281:
-------------------------------------------
offline discussion:
There's two things that might work:
a) Alter the string memoization to use a different data structure that doesn't
duplicate any part of the string (we're OOMing even with the patch)
b) Switch to a different tree compare (not string based)
> Memoize toStringTree() for ASTNode
> ----------------------------------
>
> Key: HIVE-11281
> URL: https://issues.apache.org/jira/browse/HIVE-11281
> Project: Hive
> Issue Type: Bug
> Components: Query Planning
> Affects Versions: 1.0.0, 1.2.0, 1.1.0
> Reporter: Ashutosh Chauhan
> Assignee: Ashutosh Chauhan
> Attachments: HIVE-11281.2.patch, HIVE-11281.3.patch, HIVE-11281.patch
>
>
> Currently it defaults to BaseTree::toStringTree() which always computes it
> recursively. For very large trees it becomes expensive. On one particular
> test query compilation time went down from 55 seconds to 2 seconds.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
