Rake tasks

[kjvarga]

I like the idea of this gem and I loathe having full Rails apps in my test suite so I thought I would give it a go testing my sitemap_generator gem. I want to test that the rake tasks are working properly and am having some issues. There is no 'environment' task and for some reason my gem's rake tasks which are supposed to be loaded by Railtie, are not being loaded.

Do you have any plans to add support for running rake tasks or have any pointers?

[kjvarga]

Ah nevermind I just figured it out. I just had to call Combustion::Application.load_tasks in my Rakefile. Now I get all the Rails rake tasks too...pretty neat.

For others not sure how the Rakefile should look, here is mine:

require 'bundler/setup' 
Bundler.require
Combustion.initialize!
Combustion::Application.load_tasks

desc 'Default: run spec tests.'
task :default => :spec

require "rspec/core/rake_task"
RSpec::Core::RakeTask.new(:spec) do |spec|
  spec.pattern = Dir.glob(['spec/sitemap_generator/**/*'])
  spec.rspec_opts = ['--backtrace']
end

I just ran my gem rake task and it generated sitemaps under spec/internal/ like you would expect. Neat!

[pat]

Great to hear you got it figured out - although, I'd recommend having no business logic in your rake task files anyway - pull it out into separate classes which you can test whenever possible.

Granted, I'm not particularly good at sticking to that rule myself.

[ndbroadbent]

We've also been having some issues with Rake tasks. Our goal is to be able to run both rake spec and rspec spec/.

The issue was that Combustion.initialize! was being called twice - in Rakefile, and also spec_helper. Since rake creates a separate rspec process, this causes rake spec to fail with the following error:

PGError: ERROR:  database "fat_free_crm_test" is being accessed by other users (ActiveRecord::StatementInvalid)
DETAIL:  There are 1 other session(s) using the database.

rake spec runs db:schema:load as a prerequisite by default, but it would crash unless Combustion.initialize! was called in the Rakefile, and Combustion handles the database reset anyway.
So I removed RSpec's default prerequisites. The final step was wrapping Combustion.initialize! in an :environment task, so that it's only called when necessary (i.e. for db:migrate).

So, now our Rakefile looks like this:

begin
  require 'bundler/setup'
rescue LoadError
  puts 'You must `gem install bundler` and `bundle install` to run rake tasks'
end
Bundler.require :default, :development

task :environment do
  Combustion.initialize!
end
Combustion::Application.load_tasks

require "rspec/core/rake_task"
RSpec::Core::RakeTask.new(:spec) do |spec|
  spec.pattern = Dir.glob(['spec/sitemap_generator/**/*'])
  spec.rspec_opts = ['--backtrace']
end

# Let Combustion handle database preparation
Rake::Task["spec"].prerequisites.clear

desc 'Default: run spec tests.'
task :default => :spec

This seems to be working OK, but not sure if it's the 'correct' way to do it...

[mcmire]

I didn't like how Combustion.initialize! tries to set up the database and was getting "duplicate migration" errors. Additionally, I needed a way to generate a migration (as that functionality is in generators and is not included in the Rake tasks that ActiveRecord provides). With that in mind, here's what my Rakefile looks like:

# === boot ===

begin
  require "bundler/setup"
rescue LoadError
  puts 'You must `gem install bundler` and `bundle install` to run rake tasks'
end

# === application ===

require "rails"
require "combustion"
require "active_record/railtie"

Bundler.require :default, Rails.env

Combustion::Application.configure_for_combustion

# === Rakefile ===

task :environment do
  Combustion::Application.initialize!

  # Reset migrations paths so we can keep the migrations in the project root,
  # not the Rails root
  migrations_paths = ["db/migrate"]
  ActiveRecord::Tasks::DatabaseTasks.migrations_paths = migrations_paths
  ActiveRecord::Migrator.migrations_paths = migrations_paths
end

require "rspec/core/rake_task"

Combustion::Application.load_tasks

task(:default).clear
task(:spec).clear

RSpec::Core::RakeTask.new(:spec) do |t|
  t.verbose = false
end

task default: [:spec]

#=== Additions ===

# Since we can't use `rails g migration` here, add a custom Rake task that
# gives us the ability to create a migration
# (This is cribbed from sinatra-activerecord)

require "active_support/core_ext/string/strip"
require "active_support/core_ext/string/inflections"
require "pathname"
require "fileutils"

namespace :db do
  desc "Create a migration (parameters: NAME, VERSION)"
  task :create_migration do
    unless ENV["NAME"]
      puts "No NAME specified. Example usage: `rake db:create_migration NAME=create_users`"
      exit
    end

    name    = ENV["NAME"]
    version = ENV["VERSION"] || Time.now.utc.strftime("%Y%m%d%H%M%S")

    ActiveRecord::Migrator.migrations_paths.each do |directory|
      next unless File.exist?(directory)
      migration_files = Pathname(directory).children
      if duplicate = migration_files.find { |path| path.basename.to_s.include?(name) }
        puts "Another migration is already named \"#{name}\": #{duplicate}."
        exit
      end
    end

    filename = "#{version}_#{name}.rb"
    dirname  = ActiveRecord::Migrator.migrations_path
    path     = File.join(dirname, filename)

    FileUtils.mkdir_p(dirname)
    File.write path, <<-MIGRATION.strip_heredoc
      class #{name.camelize} < ActiveRecord::Migration
        def change
        end
      end
    MIGRATION

    puts path
  end
end

[voltechs]

For those of you who might be using your engine as a git submodule in a bigger project, here is what I've settled on (riffed a bit from @mcmire's Rakefile above). I will admit that a bit of the final result is due to trail/error. For instance, I don't really know what the task :environment block does, but I know when I run rake db:migrate, it executes that block, and thus requires additional requires. *If someone else has a clearer understanding of how everything is fitting together, please let me know and I'll update`

We do not need to modify db paths in the Rakefile because we've set those up as initializers in the engine.rb so that the main app pulls migrations and schema.rb from the core project. (Bonus points for loading FactoryBot factories to be reused in the main app!)

https://gist.github.com/voltechs/49711a2f44fbb9d0df45962dfa3ce9dd

# engine.rb
module Core
  class Engine < ::Rails::Engine
    initializer :append_paths do |app|
      app.config.paths['db'] = config.paths['db'].expanded
      app.config.paths['db/migrate'] = config.paths['db/migrate'].expanded

      # So we can use core factories in the main app
      if defined?(FactoryBot)
        FactoryBot.definition_file_paths.unshift(Core.root.join('spec', 'factories'))
      end
    end
  end
end

# Rakefile
# === boot ===

begin
  require "bundler/setup"
rescue LoadError
  puts 'You must `gem install bundler` and `bundle install` to run rake tasks'
end

# === application ===

require "rails"
require "combustion"
require "active_record/railtie"

Bundler.require :default, Rails.env

Combustion.path = '/'
Combustion::Application.configure_for_combustion

# === Rakefile ===

task :environment do
  require 'dotenv/rails'
  require "action_mailer/railtie"
end

require "rspec/core/rake_task"

Combustion::Application.load_tasks

task(:default).clear
task(:spec).clear

RSpec::Core::RakeTask.new(:spec) do |t|
  t.verbose = false
end

task default: [:spec]