Contents

Setting up for DI is very similar to the setup for a service locator, but instead of passing the locator (we'll call it a _registry_ now), we only pass (or set) the dependencies that the service itself needs.

{{{lang=ruby,number=true,caption=Dependency injection example
require 'needle'

def create_application
  registry = Needle::Registry.define do |b|
    b.view          { View.new }
    b.logger        { Logger.new }
    b.database      { Database.new( b.logger ) }
    b.authenticator { Authenticator.new(b.logger, b.database) }
    b.session       { Session.new(b.logger, b.database) }

    b.app do
      app = Application.new
      app.logger = b.logger
      app.view = b.view
      app.database = b.database
      app.authenticator = b.authenticator
      app.session = b.session
      app
    end
  end

  registry[:app]
end

class Application
  attr_writer :view, :logger, :database, :authenticator, :session
end

class Session
  def initialize( logger, database )
    @database = database
    @logger = logger
  end
end

...
}}}

The @create_application@ method is now (necessarily) a little more complex, since it now contains all of the initialization logic for each service in the application. However, look how much simpler this made the other classes, especially the @Application@ class.

Now, each class no longer even needs to care that it is being initialized via another container. All it knows is that when it is created, it will be given each of its dependencies (either as constructor parameters or as property accessors).

Version data entries

2 entries across 2 versions & 1 rubygems

Version	Path
needle-1.2.1	doc/manual/parts/04_setup.txt
needle-1.3.0	doc/manual/parts/04_setup.txt