From: Brian Mitchell <binary42@...>
Date: 2006-11-28T01:34:58+09:00
Subject: Re: [ ruby-Bugs-6820 ] #ancestors never includes the singleton class (inconsistent)

On 11/27/06, Sylvain Joyeux <sylvain.joyeux@m4x.org> wrote:
> > It is supposed to.  Singleton classes (or eigenclasses, if you want to
> > call them) are not real classes, they are placeholders for per object
> > attributes, so that they are not covered by #ancestors.
> >
> >                                                       matz.
> And, out of curiosity, why do you want that ? The singleton classes look
> like the "class of a single object" in (almost) every way. In 1.9 you
> event made it inherit the "real" class of the object. So, why keeping this
> kind of reminder that we can't manipulate singleton classes like any other
> class ?
>
> (btw, I'm posting that because I had to implement my own ancestors-like
> method ...)

I can't speak for matz, but I have heard a number of reasons this is
currently so:

1) Matz has stated that the way singleton classes might change.
Details are sketchy but I think it at least means we are trying to
keep Ruby's options open.

2) You could think of all objects already having a singleton class
(not from the implementation perspective). Yet, we don't want to force
the implementation to create such a class when ancestors is called.

3) It seems easy enough to create a specific accessor since all
objects (except a few) are subject to this:

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

Of course, that might seem like a gross abuse to some and a wonderful
tool to others.

I'm not really moved one direction or another. I can say that I enjoy
the utility of being able to easily access the singleton class.

Brian.