From: Clemens Hintze <c.hintze@...>
Date: 2000-12-29T22:10:02+09:00
Subject: [ruby-talk:8255] Re: what is meant by ruby's closure?

>>>>> "Phil" == Phil Mitchell <philmi@pliocene.cenozoic> writes:

    Phil> The ruby language comparison page emphasizes that Ruby has
    Phil> better closure feature than python. What does this mean? I
    Phil> know what closure is in the mathematical sense, but for a
    Phil> programming language...?

A closure is simply some nameless code wrapped together that happens
to remember the environment/context it was created in; it is
regardless whether the context is already left!

Please look at this:

  def closure_gen(value)
    getter = lambda { value }
    setter = lambda { |newval| value = newval }
    return getter, setter
  end

  clos1get, clos1set = closure_gen(12)
  clos2get, clos2set = closure_gen(14)
  p clos1get.call    # => 12
  p clos2get.call    # => 14
  clos1set.call(15)
  p clos1get.call    # => 15
  p clos2get.call    # => 14
  
You can see here that in the context of method closure_gen there are
three method-local variables created: 

1. value (the argument to the method)
2. getter (reference to closure simply returning the value of 'value')
3. setter (reference to closure setting 'value' to a new value

The closures referred by getter and setter are returned to the
caller. The context of closure_gen is left but still alive as the
closures referred by getter and setter remembering it. As the closures
getter and setter are created in the same context, they also refer to
the same context during theirs whole lifetime.

Every call to closure_gen will incarnate a new context with three
variables value, getter and setter!

HTH,
\cle