[
https://issues.apache.org/jira/browse/HIVE-22079?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16928219#comment-16928219
]
Ashutosh Chauhan commented on HIVE-22079:
-----------------------------------------
it will be great to see perf impact of this. Any measurements on any query
before and after?
Also consider adding jmh bechmarks in tests.
> Post order walker for iterating over expression tree
> ----------------------------------------------------
>
> Key: HIVE-22079
> URL: https://issues.apache.org/jira/browse/HIVE-22079
> Project: Hive
> Issue Type: Improvement
> Components: Logical Optimizer, Physical Optimizer
> Affects Versions: 4.0.0
> Reporter: Vineet Garg
> Assignee: Vineet Garg
> Priority: Major
> Attachments: HIVE-22079.1.patch, HIVE-22079.2.patch,
> HIVE-22079.3.patch, HIVE-22079.4.patch
>
>
> Current {{DefaultGraphWalker}} is used to iterate over an expression tree.
> This walker uses hash map to keep track of visited/processed nodes. If an
> expression tree is large this adds significant overhead due to map lookup.
> For an expression trees we can instead use post order traversal and avoid
> using map.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
