Database-driven applications often need data pre-seeded into the system, such as locations or user roles. For a long time there was no good way to do this in Ruby on Rails — the primary options were seeding data directly in migrations, having migrations call a rake task to manage the relevant records, or using an external plugin.

Rails 2.3.4 largely fixed this with the introduction of the db/seeds.rb file and corresponding rake db:seed task. The default seeds.rb included with a fresh Rails install offers the following examples:

cities = City.create([{ :name => 'Chicago' }, { :name => 'Copenhagen' }])
Mayor.create(:name => 'Daley', :city => cities.first)

Rather than scattering the seeds around dozens of files and dealing with the corresponding gotchas, there’s a single, well-known place.

seeds.rb is a wonderful thing, provided nobody ever edits it, or you’re happy resetting your database every time it is edited — otherwise it’ll blindly try to recreate records every time it’s run, regardless of whether the data is already in the database. Naturally, this becomes a significant annoyance in development.

After a couple months of wondering how my coworkers and I could update the seeds file while avoiding frequent resets, I think I’ve got a straightforward solution: replace calls like

User.create(:email => "jsmith@example.com", :name => "Jane Smith")
User.create(:email => "jsmythe@example.org", :name => "John Smythe")

with

[
    { :email => "jsmith@example.com", :name => "Jane Smith" },
    { :email => "jsmythe@example.org", :name => "John Smythe" }
].each do |user|
    User.find_or_create_by_email(user)
end

written by Jordan Cole 7 August, 02010 • Comments ⁋