在rails模型中重新定义gem方法不起作用

Question

So I have a gem that contains some activerecord models that gets used by multiple applications on our platform. We had a constant defining a list of "types" that was divergent on a couple different applications.

# in the gem
class MyModel < ActiveRecord::Base
  TYPES = ["A Type", "Another Type"]
end

Redefining this constant in the application code will generate a warning for an already initalized constant. So since it's not really constant, I decided to refactor it to a class method.

# gem
class MyModel < ActiveRecord::Base
  def self.types
    []
  end
end

# application a
class MyModel < ActiveRecord::Base
  def self.types
    ["A Type", "D Type"]
  end
end

# application b
class MyModel < ActiveRecord::Base
  def self.types
    ["B Type", "C Type"]
  end    
end

However, when accessing MyModel.types in App A or B, types is [] . This, and a few other issues, have lead me to conclude that the gem is loaded after the rails app is loaded, while my assumption was the opposite.

what would be the best way to handle a divergent case such as this?

Answer 1

Do this in your applications

Create a file in your config/initializers/ and add this code in both your apps

MyModel.instance_eval do
  def types
    ["A Type", "D Type"]
  end
end

or

MyModel.class_eval do
  def self.types
    ["A Type", "D Type"]
  end
end

Answer 2

You must require the initial class explicitly from your engine, this will ensure it is loaded. Which is sometimes not the case in development ( autoload for instance)

require YourEngine::Engine.root.join('app/models/your_namespace/my_model')

class MyModel  #no inheritance here, we just reopen an existing class, right?
   def self.types
     [] #override it as you like
   end
end