creating_ruby_objects.md

1. Creating Hash - all the options

a)

[12] pry(main)> {a: 1, b: 2}
=> {:a=>1, :b=>2}

b)

Hash.new(0) # this sets default value to 0

c)

[10] pry(main)> Hash[:a, 1, :b, 2]
=> {:a=>1, :b=>2}

How is the Hash[] implemented?

It's easy!

class Hash
  def self.[](*args)
    h = {}
    0.step(args.length-1, 2) do |i|
      h[args[i]] = args[i+1]
    end
    h
  end
end

2. Creating an array

a)

[1, 2, 3]

b)

[27] pry(main)> Array.new(5, 'ar')
=> ["ar", "ar", "ar", "ar", "ar"]

c)

[33] pry(main)> Array('a')
=> ["a"]

How is the Array() implemented?

It's easy!

module Kernel
  def MyOwnArray(*args)
    args
  end
end

3. Creating a class

a)

class MyOwnHash < Hash; end

b)

MyOwnHash = Class.new(Hash)

MyOwnHash = Class.new(Hash) do
  def show_me_keys
    keys
  end

  class << self
    def new(*args)
      args
    end
  end

4. Returning different class

I want my object initializer to return different class. Impossible?

class Hash
  def initialize(*args)
    return []
  end
end

[9] pry(main)> Hash.new
=> {}

[6] pry(main)> Struct.new(:login)
=> #<Class:0x007fc5fccb94f8>
[7] pry(main)> _.class == Struct
=> false

initialize is an instance method, it is run on an object after it's created. How about this?

class Hash
  def self.new(*args)
    return []
  end

  [12] pry(main)> Hash.new('a', 'b')
  => []
end

5. Cloning object

There are 2 methods to clone object: dup and clone. While both seem to have the same behaviour, they differ a bit:

[1] pry(main)> module A; end
=> nil
[2] pry(main)> a = Object.new.extend(A)
=> #<Object:0x007ff1de023848>
[3] pry(main)> a.is_a? A
=> true
[4] pry(main)> a.dup.is_a? A
=> false
[5] pry(main)> a.clone.is_a? A
=> true
[6] pry(main)>

Bonus: Subclassing a class

[122] pry(main)> MyClass = Class.new(Class)
TypeError: can't make subclass of Class