Question

I’m using my MOO project to teach myself Test Driven Design, and it’s taking me interesting places. For example, I wrote a test that said an attribute on a particular object should always return an array, so —

t = Thing.new("test")
p t.names  #-> ["test"]

t.names = nil
p t.names #-> []

The code I have for this is okay, but it doesn’t seem terribly ruby to me:

class Thing

   def initialize(names)
      self.names = names
   end

   def names=(n)
      n = [] if n.nil?
      n = [n] unless n.instance_of?(Array)

      @names = n
   end

   attr_reader :names
end

Is there a more elegant, Ruby-ish way of doing this?
(NB: if anyone wants to tell me why this is a dumb test to write, that would be interesting too…)

Answer 1

I’d like to point out that there is already a builtin method to do what you want! It’s called Array(). The question to ask yourself is: what happens to classes that are convertible to arrays (like 0..42)?

I feel that most Rubyist would expect that they’d be converted. So:

class Thing
  attr_accessor :names

  def initialize(names)
    self.names = names
  end

  def names=(values)
    @names = Array(values)
  end
end

You will get the same results, for example:

t = Thing.new("car")
t.names  #-> ["car"]

t.names = nil
t.names  #-> []

t.names = 42
t.names  #-> [42]

t.names = [1, 2, 3]
t.names #-> [1, 2, 3]

t.names = 1..3
t.names #-> [1, 2, 3]  # Is this what you want, or not?