From: jeffperforce@...
Date: 2008-03-14T22:09:54+09:00
Subject: Closures 101

class Object
   def singleton_class
      class << self
         self
      end
   end
end

data = Object.new
status = "good"
data.singleton_class.class_eval {
   define_method(:status) {
      status
   }
}
p data.status  # => "good"
status = "bad"
p data.status  # => "bad"

I have a decent amount of experience with ruby, but I have somehow
misunderstood closures this whole time.  I thought a closure contained
a full "snapshot" of the stack, but actually it contains references to
the stack.  That is, I thought <<status = "bad">> would have no
effect.

This came about from the following situation:

def add_status(array)
   index = 0
   while index < array.size
      status = "running #{index}"
      array[index].singleton_class.class_eval {
         define_method(:status) {
            status
         }
      }
      index += 1
   end
end

array = %w(a b c)
add_status(array)
array.each { |elem| p elem.status }
# => "running 2"
# => "running 2"
# => "running 2"

If we replace "while index < array.size" with "array.size.times do"
then we get
# => "running 0"
# => "running 1"
# => "running 2"

So "times" uses a closure which in turn creates a new "status"
variable, whereas the while loop does not use a closure and there is
only one "status".

I haven't encountered this behavior before since I rarely use such
uncouth imperative-style iterations.  I'm not sure I have a point, per
se, except that it's possible to use ruby for a long time without
fully understanding closures.