[
https://issues.apache.org/jira/browse/HIVE-12734?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16186409#comment-16186409
]
Mithun Radhakrishnan commented on HIVE-12734:
---------------------------------------------
I just committed these to {{master}}, {{branch-2}}, and {{branch-2.2}}. Thank
you for the contribution, [~cdrome].
This is a performance optimization for the {{HCatLoader}}/Pig path. This should
work behind the scenes, and as such, shouldn't entail a documentation update,
IMHO.
> Remove redundancy in HiveConfs serialized to UDFContext
> -------------------------------------------------------
>
> Key: HIVE-12734
> URL: https://issues.apache.org/jira/browse/HIVE-12734
> Project: Hive
> Issue Type: Bug
> Affects Versions: 1.2.1, 2.0.0, 2.2.0, 3.0.0
> Reporter: Mithun Radhakrishnan
> Assignee: Chris Drome
> Attachments: HIVE-12734.1.patch, HIVE-12734.2-branch-2.2.patch,
> HIVE-12734.2-branch-2.patch, HIVE-12734.2.patch
>
>
> {{HCatLoader}} lands up serializing one {{HiveConf}} instance per
> table-alias, to Pig's {{UDFContext}}. This lands up bloating the
> {{UDFContext}}.
> To reduce the footprint, it makes sense to serialize a default-constructed
> {{HiveConf}} once, and one "diff" per {{HCatLoader}}. This should reduce the
> time taken to kick off jobs from {{pig -useHCatalog}} scripts.
> (Note_to_self: YHIVE-540).
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
