What's the difference between to_a and to_ary in Ruby?

Ruby to_a and to_ary method comparison
| 4 min.

Have you ever thought about the to_a and to_ary methods for the Array class? Are they the same? Why do we have two similar methods? What is the difference? What about implementations of these methods in classes other than Array? In this article I want to answer these questions. So let’s get started!

Let’s start with the definition of the to_a and to_ary methods in the Array class. On the Ruby Doc website we can see that

to_a - When self is an instance of Array, returns self. Otherwise, returns a new Array containing the elements of self.

to_ary - Returns self.

The definitions are similar, but we see the difference. When we talk about the implementation of these methods in classes other than Array, we can say that to_ary is used for implicit conversions, while to_a is used for explicit conversions. This can be hard to understand. That’s why I like to say that to_ary allows an object to be treated as an array, while to_a actually tries to convert the object into an array. This pattern is used all over the Ruby language, for example in method pairs like to_s and to_str, to_i and to_int.

OK, let’s check this out with examples. Let’s create a Point class that we can check when the to_a or to_ary method is called.

class Point
  def initialize(x, y)
    @x = x
    @y = y
  end

  def to_a
    puts 'to_a method called'
    [x, y]
  end

  def to_ary
    puts 'to_ary method called'
    [x, y]
  end

  def inspect
    "#<#{self.class.name} (#{x}, #{y})>"
  end

  private

  attr_reader :x, :y
end

Now we can check what happened when we used the splat operator (*) on the Point object.

point = Point.new(1, 3)
#  => #<Point (1, 3)>

Point.new(*point)
# to_a method called
#  => #<Point (1, 3)>

As we can see, the to_a method has been called. On another side, when we assign a Point object into variables, method to_ary is called.

x, y = point
# to_ary method called
#  => #<Point (1, 3)>

As a next step, we will check what happened when we use each on points collection.

[point].each { |item| item }
#  => [#<Point (1, 3)>]

[point].each { |(x, y)| [x, y] }
# to_ary method called
#  => [#<Point (1, 3)>]

In the first example, nothing happened. We iterated through the collection in the normal way. In the second example we see that the method to_ary is called. This is because we’re doing variable assignments. It’s worth mentioning that in the implementation of the Point class, the point components are private, but thanks to the to_ary method used in each (or another iterator), we can get access to them.

Some time ago, in the article How Ruby casts object into a string?, I described how the puts method works on the array. Now let’s see which method to_a or to_ary is used to print the array.

puts point
# to_ary method called
# 1
# 3
#  => nil

If the to_ary method is implemented in the Point class, the puts method will try to call it to split the object into smaller pieces. Then puts is called on each element. So we can say that in this case puts is called recursively.

Now is a good time to mention that if your object doesn’t have the to_ary method implemented, variable assignment will behave differently without exception. See below:

class Point
  def initialize(x, y)
    @x = x
    @y = y
  end

  def to_a
    puts 'to_a method called'
    [x, y]
  end

  def inspect
    "#<#{self.class.name} (#{x}, #{y})>"
  end

  private

  attr_reader :x, :y
end

point = Point.new(1, 3)
#  => #<Point (1, 3)>

x, y = point
#  => #<Point (1, 3)>

x
#  => #<Point (1, 3)>

y
#  => nil

So you see that the Point object has not been split into its coordinates. Now let’s check the puts method for Point without the to_ary method.

point = Point.new(1, 3)
#  => #<Point (1, 3)>

puts point
#  #<Point:0x00007f6e92b2a8b0>
#  => nil

puts does not iterate over point coordinates. A similar observation can be made for the flatten method. If the Point class does not implement the to_ary method, we will see:

point_1 = Point.new(1, 3)
#  => #<Point (1, 3)>

point_2 = Point.new(1, 5)
#  => #<Point (1, 5)>

[point_1, point_2].flatten
#  => [#<Point (1, 3)>, #<Point (1, 5)>]

On the other hand, for the Point class with the to_ary method, the result will be different.

point_1 = Point.new(1, 3)
#  => #<Point (1, 3)>

point_2 = Point.new(1, 5)
#  => #<Point (1, 5)>

[point_1, point_2].flatten
#  to_ary method called
#  to_ary method called
#  => [1, 3, 1, 5]

flatten will call to_ary (recursively) on each of the elements in the array. As an exercise, you can check the same behavior for the join method.

Summary

  1. Two methods allow us to be like an Array:
    • to_a method - allows to cast object into an Array.
    • to_ary method - allows the object to behave like an Array.
  2. Splat operator * uses to_a method.
  3. Object assignment to variables tries to use to_ary if it’s implemented. If no to_ary method is implemented, only the first variable will get the assignment. The rest of the variables will be nil.
  4. The puts method tries to use the to_ary method if it is implemented.
  5. The flatten and join methods behave similarly to puts. They try to use the to_ary method under the hood when it is implemented.

Tagged with