Why aren't these methods resolving?

Question

Given this code:

class Something
  attr_accessor :my_variable

  def initialize
    @my_variable = 0
  end

  def foo
    my_variable = my_variable + 3
  end
end

s = Something.new
s.foo

I get this error:

test.rb:9:in `foo': undefined method `+' for nil:NilClass (NoMethodError)
    from test.rb:14:in `<main>'

If attr_accessor creates a method called my_variable (and ..=), why can't foo find the method? It works if I change it to self.my_variable, but why? Isn't self the default receiver?

Answer 1

You're not calling the method there, you're actually referencing the same variable you're in the process of defining! This is a little gotcha in Ruby.

What would be better is if you referenced and set the instance variable instead:

@my_variable = @my_variable + 3

Or shorter:

@my_variable += 3

Or you could call the setter method, as you found (and Jits pointed) out:

self.my_variable += 3

This last one will call the my_variable= method defined by the attr_accessor, where the other two will only modify a variable. If you did it this way, you could override my_variable= to do something different to the value passed in:

def my_variable=(value)
  # do something here
  @my_variable = value
end

BONUS

Or you could call the method explicitly by passing an empty set of arguments through:

my_variable = my_variable() + 3

This is not "The Ruby Way" to go about it, but it's still interesting to know that you can still call a method this way if you have a local variable of the same name.