From: wys@... (Clemens Wyss)
Date: 2001-02-08T19:40:02+09:00
Subject: [ruby-talk:10559] Re: Giving a Proc utility methods?

IMHO: The reason for this behavior is, that the block closure "takes" its
binding upon its creation time, which in your case is the "main context"
of the program. This means you would have to define fail() in the global
scope in order to let the block code see it. The block closure object
(class Proc) is just a wrapper consisting of the code (i.e. the block)
and the binding, but it is NOT the scope of the block(-code). 

Don't know whether you like this solution better:

class Foo
  attr :err

      def initialize (&meth)
        @err = meth
      end

      def fail (msg)
        puts msg
      end
        
      def throw_err (msg)
        @err[msg, self]
      end
end

# I want to preserve this closure...
bar = "testing..."

# Example showing the callback works
my_foo = Foo.new {| msg, other | puts msg + bar}
my_foo.throw_err "Trial 1: "

# Example of what I *wanted* to do...
my_foo = Foo.new {| msg, other | other.fail (msg + bar)}
my_foo.throw_err "Trial 2: "


Clemens (the other ;-)