Rails Style Ruby Database Migrations with Sequel

If you’ve ever worked with ActiveRecord databases, you know how awesome it can be to manage changes to your database. Make some changes in an indexed migration file, and run rake db:migrate to apply all changes; your database stays up to date with each new git pull. Fortunately, there's a way to get the same functionality in regular Ruby code[^regular ruby] with a Rake task and the Sequel gem.

Before I figured this out, I was applying migrations in a very hacky way[^task-method]. I had a tasks method that was invoked when my global database class was instantiated:

def db_tasks @db.exec("alter table users add column age integer") rescue false @db.exec("alter table users add column subscriber boolean") rescue false end

Every time I needed to modify or create a table, I’d add a line to it. This can get messy the more you need to change the database schema. Enter the Sequel gem.

Sequel is a ruby gem for getting ActiveRecord-like functionality into your Ruby code. Instead of UPDATE users SET AGE=27 WHERE NAME="Nick" you can define a model and run User.find_or_create(:name => 'Nick').update(:age => 27) without having to worry if the record exists yet. Or instead of SELECT * FROM users LIMIT 1 you can run User.first. But you don't even have to use those features to get access to my favorite Sequel feature, migrations.

I have a directory set up within my project’s root db/migrations which has indexed migration files like 001_create_users_table.rb, 002_add_primary_key_to_users_table.rb, etc. The syntax for the files is pretty straight forward:

Sequel.migration do change do create_table :users do String :name, :unique => true Integer :age Boolean :subscriber, :default => false end end end

To apply this from the command line, you run sequel -m /path/to/migrations/dir /path/to/db and migrations are applied in order. I prefer the standard Rails command rake db:migrate so I created a rake tasks to wrap the command line argument. This is what my Rakefile looks like:

namespace :db do task :migrate do result = %x{sequel -m db/migrations/ -E postgres://localhost/db} puts result end task :create do result = %x{createdb db} end require 'highline/import' require_relative 'database_manager' require_relative 'settings' task :tasks do config_path = File.expand_path("../config/", __FILE__) @config = App::Settings.new(path: config_path) @db = App::DatabaseManager.new(settings: @config) @db.db_tasks end end

Now, to create the database I run rake db:create, to migrate changes I run rake db:migrate[^1], and to run any arbitrary tasks that I still keep in my database class, I can run rake db:taskswithout having to worry about them executing every time I create an instance of the class.

When adding team members to my project, this gets them up and running with the latest database and schema without issues.

Sequel is an incredibly powerful gem that I plan to take better advantage of as I migrate my app from ruby-pg style SQL statements to ActiveRecord compatible queries. I highly recommend it to anyone who wants to get comfortable with Rails methodologies before making the switch from desktop -based coding.

[¹]: The -E flag outputs verbose details of the migration, so I can confirm it worked

[^task-method]: Wrapping each schema alter query in a rescue clause prevents errors if the alter command has already been run.

[^regular ruby]: By this, I just mean ruby code that is not Rails.

Related Posts

Originally published at nick.prokes.ch.