Remove duplicate LICENSE and NOTICE files from benchmark JARs
- Dominant language
- Java
- Stars
- 3.1k
- Forks
- 1.6k
- Avg merge
- 3d 12h
- Merged PRs (30d)
- 33
Description
Similar to https://github.com/apache/iceberg/issues/17406
**These local JMH artifacts are not published, so releases are unaffected**. Update their shade configuration to retain only the intended `META-INF/LICENSE` and `META-INF/NOTICE`, while preserving required attribution in those canonical files.
The shaded benchmark uber-JARs contain multiple LICENSE/NOTICE variants:
- `parquet-benchmarks.jar`
- `parquet-plugins-benchmarks.jar`
## Reproduction
```bash
# Build all default and vector-plugin JARs without running tests.
./mvnw --batch-mode -Pvector-plugins -DskipTests clean package
# Check all final JARs; exclude only Maven Shade backup artifacts.
(
failed=0
while IFS= read -r -d '' jar; do
licenses=$(jar tf "$jar" | grep -Eic '(^|/)LICENSE([._-].*)?$' || true)
notices=$(jar tf "$jar" | grep -Eic '(^|/)NOTICE([._-].*)?$' || true)
if [ "$licenses" -eq 1 ] && [ "$notices" -eq 1 ]; then
echo "PASS: $jar"
else
echo "FAIL: $jar (LICENSE=$licenses, NOTICE=$notices)"
jar tf "$jar" | grep -Ei '(^|/)(LICENSE|NOTICE)([._-].*)?$'
failed=1
fi
done < <(
# original-* JARs are temporary backups created by Maven Shade.
find . -type f -path '*/target/*.jar' \
! -name 'original-*.jar' \
-print0 | sort -z
)
exit "$failed"
)
```
Observed failures:
```text
FAIL: ./parquet-benchmarks/target/parquet-benchmarks.jar (LICENSE=9, NOTICE=4)
FAIL: ./parquet-plugins/parquet-plugins-benchmarks/target/parquet-plugins-benchmarks.jar (LICENSE=4, NOTICE=3)
```
Expected: every final JAR contains exactly one LICENSE and one NOTICE file under `META-INF/`.
Contributor guide
No contributing guide indexed for this repository
Research direction
Inspect the shade configuration in the parquet-benchmarks and parquet-plugins/parquet-plugins-benchmarks modules. Run ./mvnw --batch-mode -Pvector-plugins -DskipTests clean package, then inspect the final benchmark JARs with the provided checks. Done means each final JAR contains exactly one META-INF/LICENSE and one META-INF/NOTICE while retaining required attribution.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- build-system
- Issue type
- Bug
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 72/100