Auto-load the seed data from db/seeds.rb with rake

Question

I\'m using rails-rspec gem and I have several specs (models, controllers, etc). When I run:

bundle exec rake

everything is te

Answer 1

I set up my rake spec task to automatically load db/seeds.rb, so I depend on that for setting up the database for a first run. Add this to your Rakefile:

task :spec     => "db:seed"
task :cucumber => "db:seed"

Then, on subsequent runs I just call rspec spec directly to skip the seed step and avoid unnecessary reloading. I configure database cleaner to ignore the seed tables like this:

RSpec.configure do |config|

  config.add_setting(:seed_tables)
  config.seed_tables = %w(countries roles product_types)

  config.before(:suite) do
    DatabaseCleaner.clean_with(:truncation, except: config.seed_tables)
  end

  config.around(:each) do |example|
    if example.metadata[:no_transactions]
      DatabaseCleaner.strategy = :truncation, {except: config.seed_tables}
    else
      DatabaseCleaner.strategy = :transaction
    end
    DatabaseCleaner.start
    example.run
    DatabaseCleaner.clean
  end
end

For scenarios that need committed data, I can add:

describe "commit for real", use_transactions: false do
  # ...
end

This will truncate everything after the example runs, except the seed tables. It's assumed that you never write anything to the seed tables! This is generally a safe assumption, since seed data is typically static.

For all other normal scenarios, I depend on transactions to roll back any inserted records. The database is returned to the original state, with the data in the seed tables intact. It's safe to write to the seed tables here if you need to.

To rebuild the seed tables, you just need to run rake spec again.