Brut RB

Appearance

See Data for Development ​

In Brut, seed data is data you set up for the purposes of doing local development. It is not for managing production data and is not used for tests.

Overview ​

Seed data lives in app/src/back_end/data_models/seed. You can create as many files in here as you like. Each should contain a class that extends Brut::BackEnd::SeedData and implements seed!. By doing this, the class is regsitered with Brut and when you run bin/db seed all the classes are created and seed! is called.

FactoryBot will be set up for you, so you can call FactoryBot.create to create the data.

For example, here is how we might create two accounts, one deactivated, in the same organization:

ruby
# app/src/back_end/data_models/seed/all_seed_data.rb
class AllSeedData < Brut::BackEnd::SeedData
  include FactoryBot::Syntax::Methods

  def seed!
    organization     = create(:organization)
    active_account   = create(:account,
                              organization:,
                              email: "pat@example.com")
    inactive_account = create(:account, :inactive,
                              organization:,
                              email: "chris@example.com")
  end
end

You can store other data here, such as a CSV, if you want to load data organized in another way. Seed data will only ever be loaded in development, so you can organize it however you like.

Testing ​

You do not need to test your seed data. Presumably, you will be using your app in development, and this will be sufficient to determine if your seed data is fit for purpose.

  • Use FactoryBot as this is consistent with your tests
  • Use literal values for anything relevant to local development. In the example above, the two email addresses are important, presumably because you'd be using those to login. The organization name, however, is irrelevant, so we allow Faker to come up with a name.
  • Use local variables as documentation. In the example above, there's no need to set active_account or inactive_account, but doing so makes it clear what those objects are for.
  • Ensure all seed data classes are independent from one another, as the order of their exeuction cannot be guaranteed.

Technical Notes ​

IMPORTANT

Technical Notes are for deeper understanding and debugging. While we will try to keep them up-to-date with changes to Brut's internals, the source code is always more correct.

Last Updated May 8, 2025

All seed data is loaded in one transaction. This means that if any class' seed data fails, no data will be written.

Seed data also is not assumed to be idempotent. If you run it twice, you will likely get an error. Because your dev database is ephemeral, you can always recreate your dev database via bin/db rebuild && bin/db seed.