[
https://issues.apache.org/jira/browse/LUCENE-6133?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14258099#comment-14258099
]
Adrien Grand commented on LUCENE-6133:
--------------------------------------
+1 I'm wondering if we should bring the reuse of the byte[] to DataOutput in
general?
> improve default stored fields merge algorithm
> ---------------------------------------------
>
> Key: LUCENE-6133
> URL: https://issues.apache.org/jira/browse/LUCENE-6133
> Project: Lucene - Core
> Issue Type: Task
> Reporter: Robert Muir
> Attachments: LUCENE-6133.patch
>
>
> This is not very efficient today.
> We should try to improve the default one and not rely so much upon bulk
> merges. At the least we can use a specialized visitor, save some cpu, and
> avoid creating so much garbage.
> Another pain point in default merge is all the string creation and unicode
> encode/decoding back and forth. I don't tackle that here... but that would be
> fantastic to fix in another issue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
