From: "ara.t.howard" <ara.t.howard@...>
Date: 2009-09-30T06:18:59+09:00
Subject: [ruby-core:25865] struggling to convince myself 1.9's constant lookup rules make any 	sense


i have the following issue in a library, here distilled minmally:



cfp:~/src/git/tagz > cat a.rb
module Tagz
  def Tagz.singleton_class(&block)
    @singleton_class ||= (
      class << Tagz
        self
      end
    )
    block ? @singleton_class.module_eval(&block) : @singleton_class
  end

  module XChar
    EscapeProc = lambda{|*values| 42 }
    Tagz.singleton_class{ define_method(:escape_proc){ EscapeProc } }
  end
end

p Tagz.escape_proc()


cfp:~/src/git/tagz > ruby18 a.rb
#<Proc:0x00034274@a.rb:12>


cfp:~/src/git/tagz > ruby19 a.rb
a.rb:13:in `block (2 levels) in <module:XChar>': uninitialized
constant Module::EscapeProc (NameError)
        from a.rb:17:in `<main>'


and i understand that constant lookup in 1.9 is different - occuring
in the caller's dynamic lookup chain - but i am really struggling to
see how this behavior can be called correct with respect to closures.

in the above code changing from a constant to a variable - or fully
scoping the contant - are both obvious fixes.  however i don't
understand how this new behavior isn't at odds with the fundamental
concept of 'closure': can a closure which only encloses *some* of it's
state really be called a closure?

regards.