I\'m trying to learn the life cycle of a rails application. When is application_controller.rb run? Is it just once every time it\'s changed, or on every request?
I want
With a little effort, you can follow it through yourself, which will probably be more useful.
Start from 'ruby script/server'. In my (2.1) application, that look for a file named "server" in the "script" directory. It contains this:
#!/usr/bin/env ruby
require File.dirname(__FILE__) + '/../config/boot'
require 'commands/server'
So it requires boot.rb, which defines a whole lot of stuff and then calls Rails.boot!
which more or less runs any pre-initialization you've defined and then requires environment
, which does another level of bootstrapping. By that time it's starting to get complicated: I'd recommend a large-ish sheet of paper...
And so on.
Alternatively, you may be able to hack Kernel#require
to log when a file is being required in - I haven't tried it myself (and the method may be overridden elsewhere) but it could work...
application_controller isn't exactly "run" at all - it's a parent class for your other controllers, so its content, if required and not overridden becomes available when the inheriting controller is loaded. You'd generally use it to provide common functionality across all your controllers.
It is common practice to put initialization stuff in the config/initializers/ directory. This way you can keep your environment.rb files clean(er).
See this post by Ryan Daigle.
ApplicationController is a parent class to all controllers. Methods declared in it will be available to all controllers for this reason.
ApplicationController is a convenient place to filters that you want to apply to all controllers in your application, or methods that you wish to make available to all of them.
The files in config/environments/*.rb override settings in the default config/enviornment.rb file depending on what environment your server is running in (development/production). One example is that in development errors are printed to the screen and in production a generic error page is returned. This setting is in config/environments/development.rb
boot.rb is used as part of the rails initialization process. You usually don't need to, and likely shouldn't touch it.
environment.rb is the generic configuration file for your application.
routes.rb is used to define how your application handles requests to specific urls. For example, you may want to have all 404 requests go to a specific action instead of being handled by the default error handler:
map.connect '*path', :controller => 'home', :action => 'on_404'
It is also an important part of implementing a RESTful application.
Both initialization code and custom configuration data should be placed in enviornment.rb (read the comments in this file). If you want certain code to run during initialization only in development or only in production, place it in config/environments/development.rb or config/environments/production.rb respectively.
Edit:
A good overview on when each of these files is run during initialization is available here:
http://toolmantim.com/articles/environments_and_the_rails_initialisation_process https://github.com/toolmantim/toolmantim/blob/master/articles/environments_and_the_rails_initialisation_process.haml
Essentially the steps are:
The Rails Initializer is loaded (http://api.rubyonrails.org/classes/Rails/Initializer.html)
The rails Initializer sets up logging and then loads environment.rb
environment.rb loads boot.rb
boot.rb sets the RAILS_ROOT
constant and adds rails libraries and application code to the LOAD_PATH
environment.rb executes Rails::Initializer.run
.
The rails framework is loaded (ActiveRecord, ActionMailer, etc.)
Your environment's specific config file is loaded (config/environments/development.rb.)
after_initialize
and to_prepare
callbacks are executed if you have created any
Rails has finished loading and is ready to handle requests
Here is what I observed by putting a bunch of print statements in the code:
The application_controller and the posts_controller (for example) are "run" on each request. Every line of code not inside a method def/end is executed in both controllers, and then code flow passes to the requested/routed method.
As this is not what I expected I thought it worthwhile to post it. If I'm wrong pls feel free to edit.
With Heroku, print statement output only shows up in the stdout log if the print statement is inside a method, for some reason. But I believe the above description still applies.