From: zuzu <sean.zuzu@...>
Date: 2004-08-29T10:41:22+09:00
Subject: Re: accessors break "no state change outside the object"?

On Sun, 29 Aug 2004 10:20:32 +0900, Florian Gross <flgr@ccan.de> wrote:
> zuzu wrote:
> 
> > ruby accessors are merely shortcuts for reading/writing @-scope values
> > in an object.
> > http://www.ruby-doc.org/docs/ruby-doc-bundle/UsersGuide/rg/accessors.html
> 
> As I understand it Ruby's accessors are just convenient ways of adding
> something to the interface of your Object that does modify state
> internally. (Without the interface part enforcing that.)
> 
> attr_accessor basically does this:
> 
> class Module
>    def attr_accessor(name)
>      define_method(name) do
>        instance_variable_get(:"@#{name}")
>      end
>      define_method(:"#{name}=") do |value|
>        instance_variable_set(:"@#{name}", value)
>      end
>    end
> end
> 
> That is, attr_accessor :foo creates two methods foo() and foo=(). You
> can also create those manually and they need not be related to an
> instance variable at all so I don't think this breaks encapsulation.

hmm... ok.

> > why create special syntax for initialization / object creation (.new)
> > that also isn't specified with the .initialize / .new method?
> 
> I don't see how this is related, but .new / .initialize are *no* special
> syntax.

i was thinking that commonly accessors were used to modify
object-local scope values in ways that seemed like initializing tasks,
but i now am rather sure i was mistaken.  so nevermind this.

> Class#new is defined like this:
> 
> class Class
>    def new(*args, &block)
>      result = self.allocate
>      result.send(:initialize, *args, &block)
>      return result
>    end
> end
> 
> Class#allocate is special of course, because it needs to construct an
> Object. You can't easily do that manually. (It's possible when using
> evil-ruby, but there's no real purpose in doing so.)
> 
> So basically I don't see where accessors or Class#new break Ruby's basic
> principles. I think Ruby is very consistent here. But then again I might
> just not be understanding your objections -- if so, please try to
> explain them in more detail.
> 
> > peace,
> > -z
> 
> Regards,
> Florian Gross

thanks,
-z