SlothDB is an independent open-source project. It's MIT licensed, built in the open, and shaped by the people who use it. Here's how to reach out.
The fastest way to get an answer. Bug reports, install help, weird query plans, "is this slower than it should be", feature ideas - drop them in the relevant channel and the maintainer reads everything. Good for the questions that come before you file a GitHub issue.
Join the Discord server ->Bug reports, crashes, wrong results, and feature requests. The fastest way to get a fix: include a minimal reproduction, the SQL you ran, and the input file (or a synthetic version of it).
Open an issue on GitHub ->Questions, design discussion, benchmark debates, and "how do I do X" without it being a bug. Also where new features get scoped before they become issues.
Browse discussions ->For anything that does not fit in a public issue (licensing, commercial use, partnership), reach out on GitHub. Open a direct thread via the profile, or start a private discussion.
github.com/SouravRoy-ETLPlease disclose security issues privately first. Bounds-checked parsing and DoS limits are in place, but the project is young. Report via a private GitHub security advisory and we will coordinate a fix before public disclosure.
Pull requests are welcome for bug fixes, missing SQL features, new file format readers, docs, and benchmark improvements.
Good entry points: failing cases in the test suite, missing SQL functions from the standard dialect, docs improvements, new examples. Look for issues tagged good-first-issue.
git clone https://github.com/SouravRoy-ETL/slothdb cd slothdb cmake -B build -DSLOTHDB_BUILD_SHELL=ON \ -DSLOTHDB_BUILD_TESTS=ON cmake --build build --config Release ctest --test-dir build -C Release
C++20, .clang-format enforced. Avoid adding dependencies. Every perf optimization should ship with a benchmark delta in the commit message.
403 tests, 131 000+ assertions via doctest. Every new feature needs a test. Benchmarks live in real-life-testing/ (run before and after perf changes).
SlothDB is v0.2.0. About seven months old, one primary author, CI-green on Windows, Linux and macOS. Use it in real workflows, expect to file bugs. It's not production-critical yet.
The project is developed in the open. Every optimization, every benchmark change, every extension is one commit in the git log.
.ask natural-language-to-SQL sub-REPL in the CLI shell (rules-first, ~50 KB; optional local Qwen GGUF fallback under -DSLOTHDB_ASK_MODEL=ON), CREATE TABLE AS SELECT (including from files), catalog-introspection C API for Python / Node / WASM bindings, README integrity pass that retracted an earlier inflated benchmark range. A later round of cleanup is described in the homepage Status section.CREATE LIVE VIEW with incremental CSV-append path for log-tail workloads, SLOTHDB_EDGE build for sub-MB WASM bundles (Cloudflare Workers target), DESCRIBE, PRAGMA table_info / database_list, VARCHAR(n) length enforcement, ORDER BY on narrowed projections, JOIN reverse-order equi-join fix@slothdb/wasm npm package, binder / planner fixes (non-reserved keywords as identifiers, SELECT * ORDER BY, UNION + JOIN file-literal preprocessing, COPY with bare file literal)https:// and public s3:// URLs), extended DATE_TRUNC (week/quarter/decade/etc.), new scalars MONTHNAME / DAYNAME / LAST_DAY / MAKE_DATEPhysicalArrowScan and PhysicalSQLiteScanSee the full CHANGELOG for per-commit details.
SlothDB is inspired by DuckDB, SQLite, ClickHouse, and the body of academic work on vectorized query execution (Boncz, Kersten, Neumann).
Open source. Contributors listed on the GitHub contributors graph.
If you're using SlothDB in a product or research project, an email or a mention on GitHub Discussions helps keep the project going.