-
Notifications
You must be signed in to change notification settings - Fork 5
Lecture 02
You should always have on mind that with great power comes great responsibility.
In Ruby code can be separated into files. To load code from file use require
require "somefile.extension"
File with extension [rb, so, o, dll, bundle, jar] can be addressed without the extension. Having a file called somefile.rb these two are equivalent
require "somefile"
require "somefile.rb"
The file, if it is not absolute path, will be looked from paths contained in
Every line in Ruby is an expression. That means that every line of Ruby is just executable code. There is a top level context that is just simple instance of Object (required files are executed in this context).
For example:
class A
puts "Hello world"
end
will print "Hello world" and define new class A. We can put executable code anywhere in Ruby. And everything is just an execution.
Ruby behaves in a way that every code is executable - is an expression. By definition an expression should have some return value - the result of it's execution. Let's take a look at this example
class A
end
What is the result of such an expression? You might say "we defined class A". But it is not. The fact we defined a class is only an effect of the expression but not a result ... result is a value returned from the expression itself.
var = nil
puts var.inspect
# => nil
var = class A
end
puts var.inspect
# => nil
The value have not changed .... there is no resulting value. Wrong again. The resulting value was simply nil, the same value as in the variable, but it was returned.
var = nil
puts var.inspect
# => nil
var = class A
self
end
puts var.inspect
# => A
In this example we get a return value. As we can see, it is a class A. How is that possible? The last line of a method is return value. Do you remember? Basically we can generalize to
last line of an expression is it's return value.
As said before, everything in executed in some context. This context is know as current object and is always represented by self.
self.class
# => Object
class B
self
end
# => Class
class A
def call
self
end
end
A.new.call
# => #<A:some number>
Unlike most languages, Ruby classes are open for modifications. This way programmer can modify behavior of classes defined by frameworks or Ruby itself. This technique is called Monkey patching.
class Clazz
def call
"A"
end
end
class Clazz
def call
"B"
end
end
Clazz.new.call()
# => "B"
People unfamiliar with Ruby might ask, why is the precedent example valid and the answer is simple:
classes are instances of class Class
everything in Ruby is an object ... even a class. Don't you believe me? Try this example:
class A
def self.call
"called"
end
end
class B
end
def B.call
"called"
end
C = Class.new
class C
def self.call
"called"
end
end
D = Class.new
def D.call
"called"
end
A.call # => "called"
B.call # => "called"
C.call # => "called"
D.call # => "called"
In the example we defined 4 classes, with class method call that return string "called" all those 4 techniques are equivalent and you can mix them freely.
One class can inherit from another. Ruby has only single-class inheritance - you can not inherit from multiple classes.
class A
def call
"called"
end
end
class B < A
end
C = Class.new(B)
B.new.call
# => "called"
C.new.call
# => "called"
When a class needs to inherit from multiple classes, Ruby allows as to mix in multiple Modules. Methods of these modules are then available to the class.
module Methods
def call
"called"
end
end
class A
include Methods
end
A.new.call
# => "called"
Ruby let us get the name of class by calling method name of the class.
Array.name
# => "Array"
[].class.name
# => "Array"
As everything else in Ruby even methods are objects of class Method.
Ruby has powerful tools to introspect object. One of the cases is the list of methods of an objects. The list can be obtained simple by calling method methods.
class A
def call
end
end
A.new.methods
# => array of methods
What's this good for? Let's continue.
Sometimes it might be useful to pass around only a method instead of the whole object. Ruby lets you extract a method for later use.
class A
def call(arg1)
self
end
end
meth = A.new.method(:call)
# => #<Method: A#call>
This example shows, that we "extracted" method call from class A. The method is bound to the instance of class A - the method will be evaluated in the context of the object. The method can be executed by calling call method with appropriate arguments.
meth.call("some string")
# => #<A:some_number>
From the example is obvious, that the method is executed in the context of the object.
Because Ruby is very dynamic language, we can not be always sure what type of argument we receive. Usually the programmer does not care what class the argument is, but whether the argument response to some method. This is called Duck typing technique - we do not care what the object is, we only care whether it behaves as we expect.
class A
def call
end
end
a = A.new
a.respond_to?(:call)
# => true
a.respond_to?(:wtf)
# => false
class A
def call
end
end
A.new.call
This example shows how to call a method, but there is one big "but". We have to know the name of the methods beforehand ... in the time we write the code. What if we do not the method name and we need to call it. Do not be surprised, this is very common use-case in Ruby. The two call to methods call are identical.
class A
def call(arg1)
end
end
a = A.new
a.call("some string")
a.send(:call, "some string")
Well, not so identical. When you use the send method on an object, you effectively bypass the access modifiers. This way a developer is allowed to call event protected or private methods.
The way to define methods shown before is not the only one. We can also define in a more programmatically way. It makes sense. We can inspect methods of an object, we can extract methods of an object and also call methods of an object in a dynamic way.
Class.define_method is private
class A
end
a = A.new
logic = Proc.new do
"data"
end
A.send(:define_method, :some_method_name, logic)
a.some_method_name
# => "data"
What is an object? Let's simplify it ... and show you as a contrast to classes
objects define state x classes define behavior
Object are complements to classes. We define some behavior as a class, then creating an object of the class that holds some state. Every object has to be of some class.
To create an object of some class it is used the method new of respective class.
class Dog
end
dog = Dog.new
Previously we defined many methods in simple and fancy styles. But let's get back to the core and try to define a method
class A
def call
end
end
here we use def keyword to define method. Where will def define the method? The answer is simple and complex
def defines method into the nearest class
So in the previous example the nearest class is A. That is obvious from next example when we inspect the self inside the class.
var = class A; self; end
var.class
# => Class
var.name
# => "A"
A.new.call
# => "string"
Now let's try to define a class method.
class A
def self.call
"string"
end
end
Where will Ruby define the method now?? It is a bit more complicated. To understand this, we have to explain something else first.
To understand how Ruby works, we have to understand what eigenclasses are. Let's start with simple definition
every object in Ruby has it's own eigenclass => an instance of Class
Why is this important? Because, however the eigenclasses is basically invisible to Ruby, it takes an important part in lookup paths.
When Ruby looks up a method a program is calling, it follows a basic chain (we will talk about it a bit later). Important is, that before the class the object inherits from, there is the object's eigenclass. Now it is important that the closest class to an object is not it's class but it's eigenclass.
So, back to the example we were talking about
class A
def self.call
"string"
end
end
to see it more clearly we can rewrite this example identically as
class A
end
def A.call
"string"
end
these two expressions are identical. To understand why it is important to understand this
class A
end
scope = class A
self
end
A == scope
# => true
but cack to the original quieting ... where are we trying to define the method? In the context of the instance of the class A. The important part is the instance of ... what is the closest class to instance (object)? As stated above ... it's eigenclasses. From this point you should see that
there are no class methods in Ruby
What could be called a class method is only an instance method defined on the eigenclass associated with object that represents the class itself.
So eigenclass is some stealth object that we can not see? Not really. Ruby has concept to access eigenclasses
# 1.8 and 1.9
eigenclass = class << some_object
self
end
# 1.9
eigenclass = some_object.singleton_class
now that we can access eigenclasses, let's see how we could define "class methods" (instance methods of eigenclass)
class A
def self.call
"called"
end
end
class B
class << self
def call
"called"
end
end
end
class C
end
class << C
def call
"called"
end
end
D = Class.new
class << D
def call
"called"
end
end
E = Class.new
def E.call
"called"
end
F = Class.new
class F
class << self
def call
"called"
end
end
end
all those examples are identical.
Now that you know where and how are methods defined, lets see how methods are looked up. Let's see how the class hierarchy looks for class
SomeClass -> Class -> Module -> Object -> BasicObject # BasicObject only in 1.9
and for objects
object -> SomeClass -> Object -> BasicObject # BasicObject only in 1.9
and there is a nice diagram to see the lookup path. Created by Made by DNA.
Important thing about eigenclasses is that they are not visible as classes of objects.
o1 = Object.new
def o1.meth
"string"
end
o1.meth
# => "string"
o1.class
# => Object
o2 = Object.new
o2.meth
# => undefined method `meth`
o2.class
# => Object
This example shows that having two instances of same objects. Both can behave differently. Because in the case of o1 the method is stored in the eigenclass, that is not accessible by o2.
Eigenclasses are used when a specific behavior of an object is expected