apache / apache/lucene

Over synchronization for read-only index readers in SegmentTermDocs [LUCENE-2171]

Open
#3,247 4 comments 0 reactions 0 assignees View on GitHub
affects-version:2.9.1 affects-version:3.0 legacy-jira-fix-version:4.9 legacy-jira-fix-version:6.0 legacy-jira-label:dead legacy-jira-priority:Minor module:core/search type:enhancement
Dominant language
Java
Stars
3.6k
Forks
1.4k
Avg merge
2d 11h
Merged PRs (30d)
88

Description

In SegmentTermDocs constructor (from 2.9.1)

```Java
46 protected SegmentTermDocs(SegmentReader parent) {
47 this.parent = parent;
48 this.freqStream = (IndexInput) parent.core.freqStream.clone();
49 synchronized (parent) {
50 this.deletedDocs = parent.deletedDocs;
51 }
52 this.skipInterval = parent.core.getTermsReader().getSkipInterval();
53 this.maxSkipLevels = parent.core.getTermsReader().getMaxSkipLevels();
54 }
```

The synchronization on "parent" for accessing deletedDocs is unnecessary on readonly indexes. If that access was moved into the SegmentReader then it could be protected there by default and overridden in ReadonlySegmentReader.

---
Migrated from [LUCENE-2171](https://issues.apache.org/jira/browse/LUCENE-2171) by Jayson Minard, updated May 09 2016

Contributor guide

Open the contributing guide

Research direction

Start at the SegmentTermDocs constructor and trace how SegmentReader exposes deletedDocs. Compare that access with ReadonlySegmentReader, then verify that read-only index readers no longer synchronize on the parent while accessing deletedDocs and that the existing test suite still passes.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
search
Issue type
Refactor
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.