From: David Alan Black <dblack@...>
Date: 2002-07-03T10:43:59+09:00
Subject: RE: Treatment of class variables in inherited classes

Hello --

On Wed, 3 Jul 2002, Gray, Jeff wrote:

> > From: David Alan Black [mailto:dblack@candle.superlink.net]
> > > class Box
> > >   # @@stuff = {:e => 1}   # intentionally commented out
> > >   def Box.add(k, v)
> > >     @@stuff[k] = v
> > >   end
> >
> > [...]
> >
> > > Crate.add(:d, 4)
> >
> > Don't you get an "uninitialized class variable" error at that point
> > (if line 2 is commented out)?
>
> Actually, the code runs without error on our site installation
> of Ruby, which is v1.6.5.  Only Box.add elicits the error with
> line 2 commented out, as one would expect.  But on my local build
> of v1.6.7 (and 1.7.2), Crate.add does evoke the uninitialized class
> variable error.  Kinda makes my whole post moot...  D'oh!

Hmmm...  I only get the error on 1.6.7... but anyway, I don't think
your post is moot, because the thing with class vars not being/not
being the same object for subclasses still seems to be the case, and
it's good to know about.

Here's a simplified example, just for playing around with:

  # Assignments

  class A
  #  @@cvar = "assigned in A"  # commented out (see below)
  end

  class B < A
    @@cvar = "assigned in B"
  end

  class A
    @@cvar = "assigned in A"
  end

  # Inspection

  class A
    puts @@cvar  # assigned in A
  end

  class B
    puts @@cvar  # assigned in B, but A if line 2 is uncommented
  end


David

-- 
David Alan Black
home: dblack@candle.superlink.net
work: blackdav@shu.edu
Web:  http://pirate.shu.edu/~blackdav