Exporting schemas

Store all schema versions of your app for validation.

By design, drift's code generator can only see the current state of your database schema. When you change it, it can be helpful to store a snapshot of the older schema in a file. Later, drift tools can take a look at all the schema files to validate the migrations you write.

We recommend exporting the initial schema once. Afterwards, each changed schema version (that is, every time you change the schemaVersion in the database) should also be stored. This guide assumes a top-level drift_schemas/ folder in your project to store these schema files, like this:


Of course, you can also use another folder or a subfolder somewhere if that suits your workflow better.

Exporting the schema

To begin, lets create the first schema representation:

$ mkdir drift_schemas
$ dart run drift_dev schema dump lib/database/database.dart drift_schemas/

This instructs the generator to look at the database defined in lib/database/database.dart and extract its schema into the new folder.

After making a change to your database schema, you can run the command again. For instance, let's say we made a change to our tables and increased the schemaVersion to 2. To dump the new schema, just run the command again:

$ dart run drift_dev schema dump lib/database/database.dart drift_schemas/

You'll need to run this command every time you change the schema of your database and increment the schemaVersion.

Drift will name the files in the folder drift_schema_vX.json, where X is the current schemaVersion of your database. If drift is unable to extract the version from your schemaVersion getter, provide the full path explicitly:

$ dart run drift_dev schema dump lib/database/database.dart drift_schemas/drift_schema_v3.json

What now?

Having exported your schema versions into files like this, drift tools are able to generate code aware of multiple schema versions.

This enables step-by-step migrations: Drift can generate boilerplate code for every schema migration you need to write, so that you only need to fill in what has actually changed. This makes writing migrations much easier.

By knowing all schema versions, drift can also generate test code, which makes it easy to write unit tests for all your schema migrations.