squash

Squash a range of migrations into a single migration.

Why use it

• Reduce clutter when many small migrations accumulate over time.
• Combine related schema changes into one clean migration for easier review and maintenance.
• Simplify migration history before merging a feature branch.

Usage

npm

npm exec durcno squash <start> <end> [--config path/to/durcno.config.ts] [--force] [--skip-db]

pnpm

pnpm exec durcno squash <start> <end> [--config path/to/durcno.config.ts] [--force] [--skip-db]

Bun

bunx durcno squash <start> <end> [--config path/to/durcno.config.ts] [--force] [--skip-db]

Yarn

yarn durcno squash <start> <end> [--config path/to/durcno.config.ts] [--force] [--skip-db]

Arguments

• <start> — the folder name of the first migration in the range. Use durcno status to find migration identifiers.
• <end> — the folder name of the last migration in the range.

Common options

• --config <path> — Path to config file (defaults to durcno.config.ts).
• --force — Squash even if custom statements (ddl.custom(...)) exist in the migration range. Without this flag, the command will abort when custom statements are detected because they cannot be preserved during squash.
• --skip-db — Skip all database interaction. No mixed-state validation is performed and the durcno.migrations tracking table is not updated. Useful in CI environments or when the database is unavailable.

What it does

1. Replays all migrations before the range to build a baseline snapshot.
2. Replays all migrations through the range to build the target snapshot.
3. Generates a single squashed up.ts and down.ts by diffing the two snapshots.
4. Deletes all migration folders in the range.
5. Writes the squashed migration into a single folder using the start migration's timestamp as its name.
6. If the database is reachable and all migrations in the range are applied, removes the individual durcno.migrations tracking records and inserts a single record for the squashed migration.

Example

Suppose you have three migrations from iterating on a feature:

migrations/
├── 2025-10-01T10-00-00.000Z/   # created users table
├── 2025-10-02T12-30-00.000Z/   # added bio column
└── 2025-10-03T09-15-00.000Z/   # added age column

Squash them into one:

npm

npm exec durcno squash 2025-10-01T10-00-00.000Z 2025-10-03T09-15-00.000Z

pnpm

pnpm exec durcno squash 2025-10-01T10-00-00.000Z 2025-10-03T09-15-00.000Z

Bun

bunx durcno squash 2025-10-01T10-00-00.000Z 2025-10-03T09-15-00.000Z

Yarn

yarn durcno squash 2025-10-01T10-00-00.000Z 2025-10-03T09-15-00.000Z

Result:

migrations/
└── 2025-10-03T09-15-00.000Z/   # creates users table with bio and age

The squashed migration's up.ts contains the full combined effect of all three original migrations, and its down.ts reverses everything back.

warning

Do not squash migrations that have already been deployed to production or any shared environment. Other databases running those migrations will have no record of the squashed result, causing durcno migrate to fail or apply duplicate changes. Only squash migrations that exist exclusively in your local or development environment.

Important notes

• Custom statements: If any migration in the range contains ddl.custom(...) statements, the squash will abort unless you pass --force. Custom statements cannot be automatically preserved because they are opaque to the snapshot diffing process.
• Mixed applied/unapplied migrations: If the range contains a mix of applied and unapplied migrations, the command exits with an error listing which are in each bucket. Either apply all pending migrations first, or pass --skip-db to bypass the check.
• Tracking sync: When all migrations in the range are applied, the command connects to the database, deletes the individual tracking records from durcno.migrations, and inserts a single record pointing to the squashed migration. Pass --skip-db to skip this update.

Next: verify the result with durcno status.