[
https://issues.apache.org/jira/browse/HIVE-12839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15133973#comment-15133973
]
Julian Hyde commented on HIVE-12839:
------------------------------------
It's not essential, but I recommend storing RelMetadataQuery.instance() in a
variable if you are going to use it more than once in a method or rule
invocation. We plan (as part of CALCITE-604) to cache results in the
RelMetadataQuery instance, so if you keep the same instance you'll have to do
less work. You should *definitely* use the same instance when one metadata
method calls another.
Of course we'd like to cache results BETWEEN calls but I don't know how much we
can safely cache when the graph is mutating.
Also, very minor, but I have a convention of naming RelMetadataQuery variables
{{mq}}. They crop up a lot, so you don't want a long name, and a consistent
short name helps you find one if it is around in the same method.
> Upgrade Hive to Calcite 1.6
> ---------------------------
>
> Key: HIVE-12839
> URL: https://issues.apache.org/jira/browse/HIVE-12839
> Project: Hive
> Issue Type: Improvement
> Reporter: Pengcheng Xiong
> Assignee: Pengcheng Xiong
> Attachments: HIVE-12839.01.patch, HIVE-12839.02.patch,
> HIVE-12839.03.patch, HIVE-12839.04.patch
>
>
> CLEAR LIBRARY CACHE
> Upgrade Hive to Calcite 1.6.0-incubating.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
