[
https://issues.apache.org/jira/browse/LUCENE-9935?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17363578#comment-17363578
]
ASF subversion and git services commented on LUCENE-9935:
---------------------------------------------------------
Commit 6f5a413ec6260c243a8907cdd667545bd1a079e5 in lucene's branch
refs/heads/main from Nhat Nguyen
[ https://gitbox.apache.org/repos/asf?p=lucene.git;h=6f5a413 ]
LUCENE-9935: Clone term vectors reader for merges (#182)
The newly added assertion in the bulk-merge logic doesn't always hold
because we do not create a new instance of
Lucene90CompressingTermVectorsReader for merges and that reader can be
accessed in tests (as long as it happens on the same thread).
This change clones a new term vectors reader for merges.
> Bulk merges for stored fields when index sorting is enabled
> -----------------------------------------------------------
>
> Key: LUCENE-9935
> URL: https://issues.apache.org/jira/browse/LUCENE-9935
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Assignee: Nhat Nguyen
> Priority: Minor
> Fix For: 8.9, 9.0
>
> Time Spent: 5h 10m
> Remaining Estimate: 0h
>
> Today stored fields disable bulk merges entirely when index sorting is
> enabled. However when sorting by low-cardinality fields or when the index
> sort is correlated with the order in which documents get indexed, we could
> likely still have efficient bulk merges.
> For instance, if you are merging two segments that are sorted on a field that
> can only take 2 values, one could bulk merge the first half of the first
> segment, then the first half of the second segment, then the second half of
> the first segment, and finally the second half of the second segment.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
