Haybarn JDBC driver

A JDBC driver for Haybarn — a derived distribution of DuckDB ("Haybarn, powered by DuckDB"), published by Query Farm LLC.

This is a hard fork of duckdb/duckdb-java based on tag v1.5.2.1, with the Java package, Maven coordinates, JNI library name, and JDBC URL prefix rebranded to Haybarn. The vendored src/duckdb/ engine tree is re-vendored from Query-farm-haybarn/haybarn so the driver embeds the Haybarn build of DuckDB (Haybarn trust root, Haybarn extension repository URLs).

Use

Maven

<dependency>
  <groupId>farm.query.haybarn</groupId>
  <artifactId>haybarn_jdbc</artifactId>
  <version>1.5.2</version>
</dependency>

Code

import farm.query.haybarn.HaybarnDriver;       // SPI entry-point; loaded automatically
import farm.query.haybarn.DuckDBConnection;    // class names kept for source compat

try (Connection conn = DriverManager.getConnection("jdbc:haybarn::memory:");
     Statement stmt = conn.createStatement();
     ResultSet rs = stmt.executeQuery("SELECT version()")) {
    rs.next();
    System.out.println(rs.getString(1));   // v1.5.2
}

Migration from org.duckdb

The JDBC URL prefix changes to jdbc:haybarn: and the Java package moves to farm.query.haybarn. The class names themselves (DuckDBConnection, DuckDBAppender, etc.) are kept verbatim so user code that casts to them needs only an import-source change. The Driver SPI entry-point class is renamed: org.duckdb.DuckDBDriver → farm.query.haybarn.HaybarnDriver.

- import org.duckdb.DuckDBDriver;
- import org.duckdb.DuckDBConnection;
+ import farm.query.haybarn.HaybarnDriver;
+ import farm.query.haybarn.DuckDBConnection;
- conn = DriverManager.getConnection("jdbc:duckdb::memory:");
+ conn = DriverManager.getConnection("jdbc:haybarn::memory:");

Development

It's required to have a JDK installed to build. Make sure the JAVA_HOME environment variable is set.

To build the driver, run make release. This produces:

• build/release/haybarn_jdbc.jar
• build/release/haybarn_jdbc_tests.jar

Run the tests:

make test
# or, equivalently:
java -cp "build/release/haybarn_jdbc_tests.jar:build/release/haybarn_jdbc.jar" \
  farm.query.haybarn.TestDuckDBJDBC

Run a single test by name:

java -cp "build/release/haybarn_jdbc_tests.jar:build/release/haybarn_jdbc.jar" \
  farm.query.haybarn.TestDuckDBJDBC test_valid_but_local_config_throws_exception

Refresh the vendored engine

src/duckdb/ is a vendored copy of the DuckDB source, as modified by the Haybarn core fork. To roll forward to a newer Haybarn core, point a fresh checkout of Query-farm-haybarn/haybarn @ haybarn at vendor.py:

python3 vendor.py --duckdb /path/to/haybarn-checkout

Then bump the version in META-INF/MANIFEST.MF and the publisher scripts.

Scalar function usage examples

See UDF.MD.

Distribution

• Maven Central at farm.query.haybarn:haybarn_jdbc. Per-platform native libraries are bundled as classified JARs (linux-amd64, linux-arm64, osx-universal, windows-amd64, windows-arm64) plus a fat haybarn_jdbc.jar containing all of them.
• Releases are GPG-signed with the Haybarn release key (the same key used by Query-farm-haybarn/haybarn GitHub releases).
• Release tags are haybarn-v<version>; .github/workflows/haybarn-jdbc.yml triggers on them.

Attribution

Haybarn is independent of and not endorsed by the DuckDB Foundation. DuckDB is a trademark of the DuckDB Foundation. See NOTICE for the modifications this fork makes to the upstream duckdb/duckdb-java repository.