sqlmesh

Breaking changes in SQL transformations are detected using a plan and apply workflow that generates a diff classifying model updates as breaking, non-breaking, or new to control physical table updates and interval backfills.

Building incremental models and SCD Type 2 data transformations is best handled by a plan and apply workflow that supports incremental by time range, incremental by unique key, and SCD Type 2 model kinds with state-aware backfill intelligence.

Migrating an existing dbt project to SQLMesh is supported through an import process that accelerates adoption while preserving your existing workflow structure and model semantics for state-backed deployments.

Validating SQL model behavior before production promotion is achieved by enforcing correctness through audits that must return 0 rows and validating model behavior with input and output unit tests during CI/CD pipelines.

State-aware deployments for SQL transformations are needed to remove the risk of applying changes by tracking previous deploy states, ensuring that only the required physical table updates and interval backfills are materialized across development and production.