From: Austin Ziegler <halostatue@...>
Date: 2005-11-19T00:39:57+09:00
Subject: Re: Method Definitions

On 11/18/05, Daniel Schierbeck <daniel.schierbeck@gmail.com> wrote:
> David A. Black wrote:
>> On Fri, 18 Nov 2005, Daniel Schierbeck wrote:
>>> Just a little curiosity: If method definitions (`def foo; end') were
>>> to return a symbol representing the method's name (`:foo' in this
>>> case) rather than just nil, would this be possible?
>>>
>>>  class Klass
>>>    private def foo
>>>      "Klass#foo"
>>>    end
>>>
>>>    def bar
>>>      "Klass#bar"
>>>    end
>>>  end
>>>
>>>  obj = Klass.new
>>>  obj.bar -> "Klass#bar"
>>>  obj.foo -> private method `foo' called...
>> Yes; I think that's one of the main things people who advocate this
>> want to be able to do (just remembering [I think] from earlier
>> discussions).
> Could you point me to one of those discussions? I wonder why such a
> proposal was turned down.

As a big proponent of making def return something useful, it was turned
down in part because what would def return to be passed into private
with:

  class Klass
	private def self.foo
	  "Klass.foo"
	end
  end

If you simply do :foo, then you would be applying private to Klass#foo
-- which isn't what you want. But returning more than :foo isn't kosher
either since private doesn't (at this point) know what to do with it.

I *think* Matz would be at least partially in favour of this if we could
figure out something useful and lightweight to return that wouldn't
prevent the above code.

-austin
--
Austin Ziegler * halostatue@gmail.com
               * Alternate: austin@halostatue.ca