Introducing libsql Migrate

libsql Migrate is a database migration and seed management tool for libsql with configurable options, built with Node.js and available as an npm package.

Installation

The easiest method is a global installation via npm.

npm install -g libsql-migrate

The libsql-migrate command is now available for your usage.

Local Installation

Instead of installing globally, you can install locally at the project-level.

npm install --save-dev libsql-migrate

This prevents compatibility issues because it defines the version of libsql-migrate that the migrations/seeds in your repository were written for.

You can run npx libsql-migrate <command> and/or add migration commands to your package.json:

{
  "name": "mypackage",
  "version": "0.1.0",
  "scripts": {
    "migrate": "libsql-migrate latest",
    "seed": "libsql-migrate seed:run"
  }
}

Usage

First, navigate to your project's root directory

cd my/project/root

Then, create a fresh configuration file.

libsql-migrate init

# prefix with npx if you installed locally, like so:
npx libsql-migrate init

This adds a file called libsqlrc.js to your project with the following contents. Modify it to meet your project's configuration.

export default {
  development: {
    connection: {
      url: "file:local.db",
    },
  },
  production: {
    connection: {
      url: "libsql://...",
      authToken: "...",
    },
  },
};

Making a Migration

When you want to make changes to your database, create a new migration.

libsql-migrate make demo

A file with the current timestamp and the name you chose will be added to the migrations directory. This directory is ./migrations by default, but can be configured in libsqlrc.js like so:

export default {
  development: {
    connection: {
      url: "file:local.db",
    },
    migrations: {
      directory: "my_migrations_directory",
    },
  },
  // ...
};

Migration files look like this:

export async function up(client) {}

export async function down(client) {}

In the migration file, write the code that brings your schema up toward the latest version in the up() function. For example:

export async function up(client) {
  await client.execute(
    "CREATE TABLE users (id INTEGER PRIMARY KEY AUTOINCREMENT, name TEXT);",
  );
}

Now, write the code that brings your schema down (reverses those changes) in the down() function.

export async function down(client) {
  await client.execute("DROP TABLE users;");
}

Running the Next Migration

Once a migration is written, it needs to be run against the database to commit the changes it contains.

The up subcommand runs the next migration that has not yet been run.

libsql-migrate up

The up() function in the next migration file (alphabetically) in the migrations directory will be executed.

You can repeatedly run this command to keep migrating up. If you want to run all pending migrations to bring the database schema fully up-to-date, use the latest subcommand (detailed later).

Rolling Back the Latest Migration

Sometimes you need to undo the changes that you previously made. Use the down subcommand for that.

libsql-migrate down

The down() function in the most recently executed migration file will be executed.

You can repeatedly run this command to roll back the database schema further and further back.

Running All Pending Migrations

If you have multiple pending migrations and you simply want to migrate the database schema all the way up so it's fully up-to-date, use the latest subcommand.

libsql-migrate latest

The up() function for all pending migration files will be executed in series, alphabetically.

Rolling Back the Latest Batch

If you ran libsql-migrate latest and want to revert the changes it made, you can always roll back the latest batch.

libsql-migrate rollback

The down() function for all migrations in the last batch will be executed in series.

Making a Seed

If you want to fill your database with some preset data, start by generating a new seed file.

libsql-migrate seed:make demo

A file with the name you chose will be added to the seeds directory. This directory is ./seeds by default, but you can change that in libsqlrc.js like this if you wish:

export default {
  development: {
    connection: {
      url: "file:local.db",
    },
    seeds: {
      directory: "my_seeds_directory",
    },
  },
  // ...
};

Seed files look like this:

export async function seed(client) {}

Inside the file, write the code that seeds your database with preset data in the seed() function.

export async function seed(client) {
  await client.execute("DELETE FROM users;");
  await client.execute("INSERT INTO users (name) VALUES ('admin')");
}

Run a Seed

Use the seed:run subcommand with the name of the seed to run it.

libsql-migrate seed:run demo

This will execute the seed() function inside demo.js in the seeds directory.

Run Multiple Seeds

You can run multiple seeds by providing multiple names.

libsql-migrate seed:run demo demo2 demo3

This will run the demo, demo2, and demo3 seed files (in alphabetical order, no matter what order you type them).

Run All Seeds

You can run all seeds in alphabetical order by running the seed:run subcommand without any names.

libsql-migrate seed:run

That's all you need to get started with libsql Migrate. If you'd like to read more about how and why I created this tool, read on!

Why Did I Create This?

With PlanetScale eliminating their free tier, I was on the hunt for an alternative database hosting platform. Lately, I've been hearing some buzz about Turso. In addition to some other excellent features, their free tier allows for 9 GB of storage, 500 databases, 1 billion row reads per month, and 25 million row writes per month.

I was delighted to hear that Turso uses libsql under the hood, which is a fork of SQLite. I was already familiar with SQLite and used it with numerous small and medium size apps. Being able to develop locally with libsql and then seamlessly deploy to Turso (especially with the generous free tier) made trying it out a no-brainer.

I was eager to get started, so I created a new project and installed the @libsql/client npm package.

npm init -y
npm install @libsql/client

To start, I wanted to develop locally. So I created index.js and started coding.

import { createClient } from "@libsql/client";

const client = createClient({
  url: "file:local.db"
});

Now at this point, I could have just executed some initial queries to create a table and insert some default data like this:

await client.execute(`
  CREATE TABLE users (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    name TEXT
  );
`);

await client.execute(`INSERT INTO users (name) VALUES ('admin');`);

But I obviously didn't want to run this every time my index.js file ran.

My next thought was to move the creation and seed queries into a separate, standalone file that I could run just to initialize the database.

But I figured...

1. If I'm going through the effort of imitating a database migration system, and

2. I want the added safety of versioning my database schema

...I might as well go all-in on setting up proper database migration.

No ORM?

You might be thinking, "wait! Aren't you going to use Prisma or Drizzle or some other ORM-like tool?" It's true, if I used a tool like that, it would reduce or eliminate the need for database migrations in the way libsql Migrate does them. See, I find myself eventually preferring to write raw SQL in almost every project.

Mainly because, in a big enough application, queries become complex. So complex that you must use an ORM's raw query feature anyway. And for the simpler queries, I find myself starting by writing raw SQL in database administration software until the results look how I want them and then trying to figure out how to convert it to whatever specialized format the ORM I'm using at the time uses.

I like the way Knex.js does migrations. Unfortunately, there is no libsql support in Knex.js. So if I wanted to write Knex.js-style migrations, I'd have to write extra code to run them.

So I started a new repository and refactored the migration-running code into it. That side-project grew in scope for a while until it was a complete database migration system with all the features I needed. Thus, libsql Migrate was born!

Contributions Welcome

If you find a bug or would like to add a feature, do not hesitate to fork the project on GitHub and send a pull request! This project is MIT-licensed.