From: Jeff <cohen.jeff@...>
Date: 2006-12-29T10:55:11+09:00
Subject: Re: Paper on strong Object Oriented Encapsulation (Was: some subject with needless insults)

Eric Hodel wrote:
> After skimming for the Ruby keyword this paper has nothing wrong.  I
> can't evaluate the Smalltalk or Python claims, but they look accurate
> from my understanding of the two.

I agree.  And I also agree with James, that the OP's choice of
vocabulary was, shall we say, unfortunate. :-)

However, if I could change the discussion a little bit, I wonder *why*
encapsulation from subclasses is so important.  Ruby provides decent
protection from client code - usually a prudent use of attr_reader
helps protect instance variables, for example, and I think I read
recently in a future version of Ruby (2.0?) the use of :send to get at
private methods/attributes won't be possible anymore either.

But, the paper focuses (I admit after just a quick scan) only on the
lack of encapsulation an object can exert against subclasses.  That's
always been a part of my belief system, too, as I grew up on C++ and
later C#... but now I dare wonder aloud, has that really been all that
important?

For example, I used to think static typing was the only way to go,
until I discovered unit tests as a way to stop worrying about my
program's correctness.  So now I wonder if this worry about protection
from subclasses is even worth worrying about.

Maybe the paper's authors are correct, Python and Ruby don't do this
kind of encapsulation - but maybe the answer is evident by the
increasing populariy of dynamic languages like Python and Ruby that you
can write great software without needing it such protection anyway?

As I write this, I'm thinking that as soon as I send this message, I'm
going to think of a couple of great examples that will go against this
idea, and then I'll slap my forehead silly but none of you will hear
it...

Jeff