Getting instant feedback: the dbt v2-powered dbt developer experience
Alexander Bogdanowicz walks through how dbt v2 gives developers real-time feedback on their SQL instead of waiting for a run to fail.
Overview
dbt v1 workflows still rely on a slow feedback loop: write SQL, kick off a run, wait for failures, then dig through logs to understand what broke. With dbt v2, that changes. The same Rust-based engine behind v2 now powers a development experience that understands your project as you work – catching issues in real-time, generating parquet artifacts for fast metadata exploration, and giving you an updated model and column lineage as you or your agent develops in your dbt project.
What you'll take away
- 01
dbt v2 rewrites the Jinja templating engine in Rust, giving dbt actual semantic understanding of SQL grammar for the first time, not just DAG orchestration.
- 02
The headline number: dbt v2's linter runs 50 to 100 times faster than standard SQLFluff, because its semantic understanding fills gaps that normally require slow introspection queries against the warehouse.
- 03
The VS Code extension's Compare Changes feature runs a live data diff between a modified model and production, and in the demo caught an accidental 90,000-row drop from a single bad filter change.
Session speaker
Alexander Bogdanowicz
dbt Labs