Schema Lifecycle

The schema is versioned. VERSION at the repo root holds the current semver; every PG and GeoPackage database carries a schema_migrations table and a current_schema_version view baked in by sql/0-meta.sql. Together they let the project evolve the data model without losing track of what's in production.

SELECT version FROM current_schema_version;
-- → v0.1.0

• Versioning

  ---

  How VERSION, schema_migrations, and the baseline relate. Why semver, why a view, and where the source of truth lives.

  How versioning works

• Migrations

  ---

  Issue-NNN annotations, paired PG/GPKG dialects, UNRELEASED.sql, and the strict-sequential runners.

  Writing a migration

• Release Process

  ---

  What scripts/release.sh does, what the Release GitHub Action publishes, and how to consume a release.

  Cut a release

• CI & Pre-commit

  ---

  How the immutability hook is enforced both locally and on PRs, and the flake-pinned tool versions that keep CI deterministic.

  The CI gates