SpatialStrategy.makeRecipDistanceValueSource takes zScaling from half the corner-to-corner great-circle distance of the query's bounding box, which goes the short way round the globe once a geo box is wider than 180 degrees of longitude, so edge scores fall far below the documented ~0.1 (0.0082 for rect(-170,170,-10,10), 0.0028 for a 5S-5N band of all longitudes; released since 4.0.0)

Finding LFV-778Area: spatial-extras / SpatialStrategy recip-distance scoring (every strategy with a distance value source)Priority: lowImpact: silent wrong resultsNovelty: new

Finding page for LFV-778

AI-assisted finding, confirmed by automated checks, not yet triaged by a human Lucene committer. This site files nothing upstream.

What it shows

You score a geo spatial query with makeRecipDistanceValueSource, whose javadoc says zScaling is one tenth of the distance from the centre to the farthest edge, so scores run from 1 at the centre down to about 0.1 at the furthest edge. For a box wider than 180 degrees of longitude the method instead takes half the great-circle distance between the box's two opposite corners, and that distance goes the short way round the globe: for rect(-170,170,-10,10) zScaling is 1.41 instead of 17.0, and the edge point (170,0) scores 0.0082 instead of about 0.091 (for the band rect(-180,180,-5,5), 0.0028). Nothing fails; distance is over-weighted, so with FunctionScoreQuery.boostByValue a hit with inner score 10 at 170 degrees ranks below a hit with inner score 1 at 10 degrees on the band query, the reverse of the documented scaling.

The test shows that SpatialStrategy#makeRecipDistanceValueSource for the geo box rect(-170, 170, -10, 10) (wider than 180 degrees) scores the point (170, 0) on its east edge far below the documented value.

Expected vs actual

Expected: ~0.1 at the furthest edge (javadoc; 17 / 187 = 0.091, upstream tolerance 0.10 +- 0.09); actual (-ea and -da): 0.00823, because zScaling is taken from the corner-to-corner great circle, which goes the short way round the globe.

Files

Run it

# Lucene checkout at the tested commit, built with: ./gradlew --no-daemon assemble   (JDK 25)
LUCENE=/path/to/lucene; V=12.0.0-SNAPSHOT
CP=$LUCENE/lucene/core/build/libs/lucene-core-$V.jar
CP=$CP:$LUCENE/lucene/codecs/build/libs/lucene-codecs-$V.jar
CP=$CP:$LUCENE/lucene/test-framework/build/libs/lucene-test-framework-$V.jar
CP=$CP:$LUCENE/lucene/spatial-extras/build/libs/lucene-spatial-extras-$V.jar
CP=$CP:$LUCENE/lucene/spatial3d/build/libs/lucene-spatial3d-$V.jar
CP=$CP:randomizedtesting-runner-2.9.2.jar:junit-4.13.2.jar:hamcrest-3.0.jar   # test-framework dependencies (Maven Central)
CP=$CP:junit-platform-console-standalone-6.1.3.jar   # JUnit Platform console launcher (Maven Central)

javac -d classes -cp "$CP" $(find src -name '*.java')

# run 1: assertions on, seed DEADBEEF
java -ea -esa --add-modules jdk.incubator.vector --enable-native-access=ALL-UNNAMED -Dtests.seed=DEADBEEF \
     -cp "classes:$CP" org.junit.platform.console.ConsoleLauncher \
     execute --select-class org.apache.lucene.spatial.TestRecipDistanceBoxWiderThan180

The module(s) spatial-extras also need their third-party runtime jars; ./gradlew -p lucene/<module> dependencies lists them.

Gradle alternative: copy the test into the matching module's src/test/ tree of the checkout and run ./gradlew -p lucene/<module> test --tests <class> -Ptests.seed=<seed> (add -Ptests.asserts=false for a run without assertions).

Feedback

Open an issue in lucene-ai-bug-hunt. Please do not open issues or pull requests in apache/lucene about this finding unless a Lucene committer agrees it should be filed.

Snapshot of 2026-11-04. AI-assisted; not yet triaged by a human Lucene committer. Feedback: open an issue in lucene-ai-bug-hunt.