This one is all about methods in ruby. How we can dynamically call, define and remove methods. The difference between module methods, class methods, instance methods and singleton methods. We also cover how methods are also just objects and an interesting example of updating a method definition at runtime.

Dynamically Calling Methods

In ruby we can call any method on any object. In formal OO programing computer science terminology, calling a method is thought of as sending a message, where the object is the receiver. Ruby has a method named send on the Object class. So all ruby objects inherit send and here’s how we ‘call methods’ using send.

?> class Hello
?>   def say_hello(name)
?>     "Hello, #{name}"
?>   end
>> end

>> h = Hello.new
=> #<Hello:0x0000000107b127b0>
>> h.say_hello("Ruby")
=> "Hello, Ruby"

>> h.send(:say_hello, "Ruby")
=> "Hello, Ruby"

The first argument to send is the method name (as a symbol or a string) or the message that you are trying to send. The remaining arguments (including a block) are passed on to the method as parameters.

So send allows us to call methods dynamically by sending a message to the object as the receiver. Why would be want to use send instead of the regular dot obj.my_method syntax?

Well, with send the method name can be decided at runtime. The method name is a regular parameter and we can decide which method to call at the very last minute, dynamically. This is called Dynamic Dispatch.

The other reason is that with send, we call any method, including private methods. For example method_missing is actually a private method of BasicObject but we call it with send like this obj.send(:method_missing, :random_method).

We can, not only call methods dynamically, but also define methods dynamically. Let’s see how that works.

Dynamically defining methods

define_method is a method of the Module class. It takes a block which becomes the body of the new method being defined.

?> class MyClass
?>   define_method :ask_me_anything do |question|
?>     "The answer to your question, #{question}, is 42"
?>   end
>> end

>> obj = MyClass.new
>> obj.ask_me_anything("what is the meaning of life, the universe, and everything")
=> "The answer to your question, what is the meaning of life, the universe, and everything, is 42"

ask_me_anything becomes an instance method of MyClass. The reason to use define_method instead of def keyword is again the fact that we can wait to decide the name of the method at runtime. This is powerful and allows for much of the functionality in Rails, for example.

We can also undefine or remove methods like this:

?> class Hello
?>   def greet
?>     "Hello"
?>   end
>> end

>> h = Hello.new
>> h.greet
=> "Hello"

>> Hello.remove_method(:greet)
>> h.greet
(irb):9:in `<main>': undefined method `greet' for #<Hello:0x00000001032316e0> (NoMethodError)

Both undef_method and remove_method are methods on the Module class. undef_method is more drastic in that it will undefine any method, including inherited ones. remove_method will remove the method from the receiver object only.

There is one more thing I’ve wondered about, in terms of terminology around method types. We say modules have class methods and instance methods. What is meant my module class methods. And also, modules cannot be instantiated (that’s the main difference between modules and classes) so how do think of module instance methods? Let’s clear this up.

class vs. module vs. instance vs. singleton methods

What are module methods? Let’s see in IRB:

?> module Hello
?>   def self.greet
?>     "Hello there!"
?>   end
>> end

>> Hello.greet
=> "Hello there!"

Module methods are the ones defined with self. inside a module definition. They are called with the module name like Hello.greet in the example above.

Same deal with class methods. They are the ones defined with self. inside a class definition. I do see ‘class methods’ and ‘module methods’ used interchangeably when referring to methods like Hello.greet inside modules. And that’s accurate in a way.

So what about the term ‘module instance methods’?

?> module MyModule
?>   def instance_method
?>     "a class that includes this module will get this method"
?>   end
>> end

?> class MyClass
?>   include MyModule
>> end

>> obj = MyClass.new
>> obj.instance_method
=> "a class that includes this module will get this method"

When a module is included in a class, the module’s instance methods (the ones without self.) become instance methods of that class. That is all that’s meant by ‘module instance methods’. They are the ones that a class gains when it mixes in a module with include. (this post talks more about include, extend, etc.)

That clears up this comment in the Kernel docs for example.

Kernel module has some module methods and the rest are instance methods documented in Object class, which includes Kernel

Before we wrap up, another type of method is singleton methods.

Did you know that, in ruby, objects can have their own methods? A method that ‘belongs to’ an object. No other object, even of the same class, has that method. These are called singleton methods (this post of One Ruby Question is all about singletons methods and singleton classes. So I won’t repeat here).

I will leave you with one last fact about methods.

All methods in Ruby are objects of the class Method. So methods are also just objects in Ruby, like everything else. We can do interesting things like updating a method definition like this:

?> class MyClass
?>   def my_method
?>     puts "original method"
?>   end
>> end

>> obj = MyClass.new
>> obj.my_method
original method
                                          
>> original_method = obj.method(:my_method)  # save a copy of old method
=> #<Method: MyClass#my_method() (irb):2>
?> obj.define_singleton_method(:my_method) do
?>   puts "before calling old method"
?>   original_method.call
?>   puts "after calling old method"
>> end

>> obj.my_method  # we have updated the method definition
before calling old method
original method                                                   
after calling old method 

There is a lot going on here.

  • We are updating the definition of my_method dynamically to add functionality to it (as a singleton method on the object obj).
  • We are saving a copy of the original method and calling it inside the updated method definition.
  • We are not changing the contract my_method has with its callers. We are not changing the methods ‘api’ to the outside world. But the existing callers will get the added functionality for free, without doing anything different or even knowing about it.

You can imagine the use cases that this pattern would enable. This is the power of ruby (that it lends to rails).

P.S. In addition to methods, we also have lambdas (and procs) as a way of storing pieces of reusable code. While procs behave like blocks, lambdas behave like methods. We can think of lambdas as anonymous methods created with lambda or -> syntax. (lambda and proc is one of the more confusing features of Ruby. I will plan to make a future post on this topic).