Is there a better way to run a capistrano task fro

Question

I have a set of rake tasks where I need to invoke capistrano at some point. Edwin Goei's blog suggests shelling out to capistrano via "sh".

Is there a simpler way? It would seem you should be able to call the appropriate tasks programmatically. Thanks in advance.

Answer 1

For capistrano 3:

http://capistranorb.com/documentation/advanced-features/capistrano-pure-ruby/

require 'capistrano/all'

  stages = "production"
  set :application, 'my_app_name'
  set :repo_url, '[e-mail]:capistrano/capistrano.git'
  set :deploy_to, '/var/www/'
  set :stage, :production
  role :app, %w{}

  require 'capistrano/setup'
  require 'capistrano/deploy'
  Dir.glob('capistrano/tasks/*.cap').each { |r| import r }

  Capistrano::Application.invoke("production")
  Capistrano::Application.invoke("deploy")

Answer 2

Yes, Capistrano has programmatic access to the command-line components. If you want to call them from a rake task, though, you need to do a little extra work.

task :deploy
  require 'rubygems'
  require 'capistrano'
  require 'capistrano/cli'

  parameters = ["deploy"] # this is an array of the strings that come after
                          # cap on the command line. e.g., 
                          # ["deploy", "-S", "revision=1024"] gives you local var
                          # revision in your deploy.rb.

  # The following is required ONLY when you run Capistrano 2+ from Rake, 
  # because Rake adds the methods from FileUtils to Object. FileUtils includes 
  # a method called symlink which interferes with Capistrano's symlink task.
  Capistrano::Configuration::Namespaces::Namespace.class_eval { undef :symlink }

  Capistrano::CLI.parse(parameters).execute!
end

Answer 3

Jonathan, your mileage may vary by doing something like set(:shell, false) to stop capistrano running tasks in a sub-sh-shell.

Just a thought, feel free to ping me if you need a hand though.