Run DuckDB online

DuckDB is an in-process analytical database — think "SQLite for columns". Vectorised execution, full SQL (including window functions, CTEs, LATERAL joins), zero-config, single binary. Ideal for ad-hoc data exploration and query learning.

ls-judge runs the DuckDB CLI inside a Firecracker microVM. Each submission opens a fresh in-memory database; you write CREATE / INSERT / SELECT in one script and read the result rows.

What you get in DuckDB

Clicking "Run now" opens the playground with this exact code and a matching test case pre-loaded — passes on first Run.

SELECT 'Hello, World!';

Common questions about DuckDB on ls-judge

Which version?

DuckDB 1.1 — the first stable major line.

Persistence?

No — an in-memory DB per submission. Attach a file in your SQL if you want a scratch on-disk DB inside the VM.

Extensions?

The bundled DuckDB CLI ships with the core extensions. No network means you can't `INSTALL httpfs` or similar from the CLI session.

Parquet / Arrow?

Core engine supports them; no files are staged in the sandbox, so bring your data in via INSERT or a VALUES clause.

Free?

Yes.

• DuckDB documentation↗