// the find
dbt-labs/dbt
dbt enables data analysts and engineers to transform their data using the same practices that software engineers use to build applications.
This is the main branch of dbt, which now holds the Apache 2.0 source for dbt v2.0, a ground-up Rust rewrite of the tool that turns SELECT statements into tables and views in a warehouse. It is aimed at teams already running dbt on large projects who are hitting parse and compile times, or who want a single binary instead of a Python environment. The v1 Python implementation has moved to the 1.latest branch.
The warehouse adapters are split into their own crates, and adbc-record-replay records adapter traffic so tests can run without live connections, which is the right way to test code that talks to many warehouses. Artifacts are written as Parquet next to the existing manifest.json, so project metadata can be queried with ordinary tools without breaking anything that reads JSON. The repo also shows real release hygiene: changie entries per change, deny.toml files for dependency policy in the adapter crates, and CI that includes schema checks and a test repeater for flaky tests.
The README says this Apache source is not the whole product: the dbt distribution adds customizations under a separate product license, so the code here is not necessarily the binary people install. The stricter language spec is presented as a benefit, but the README does not say what it now rejects, so an existing v1 project may not move over cleanly. Windows on ARM is not supported yet, and the split between main and 1.latest is explained only by a single warning at the top of the README.