[
https://issues.apache.org/jira/browse/LUCENE-10126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17421376#comment-17421376
]
Nhat Nguyen commented on LUCENE-10126:
--------------------------------------
[~jpountz] Elasticsearch uses a
[chunked|https://github.com/elastic/elasticsearch/blob/dfd65beaf66968b8b3d49ee5062a885cf7f40da0/server/src/main/java/org/elasticsearch/search/internal/CancellableBulkScorer.java#L42-L47]
bulk scorer so it can cancel queries quickly.
> CompetitiveIterator of NumericComparator can wrongly skip documents
> -------------------------------------------------------------------
>
> Key: LUCENE-10126
> URL: https://issues.apache.org/jira/browse/LUCENE-10126
> Project: Lucene - Core
> Issue Type: Bug
> Components: core/search
> Affects Versions: 8.9, 8.10
> Reporter: Nhat Nguyen
> Priority: Major
> Time Spent: 20m
> Remaining Estimate: 0h
>
> The ML team at Elastic reported that a large scroll with an Elasticsearch
> nightly build that uses Lucene 9.0 snapshot returns fewer documents than
> expected. I looked into it and found that the competitive iterator can
> wrongly skip docs with a chunked bulk scorer.
--
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
