From: Nathaniel Talbott <ntalbott@...>
Date: 2001-03-23T11:52:11+09:00
Subject: [ruby-talk:13108] Re: alias method

Kenichi Komiya [mailto:kom@mail1.accsnet.ne.jp] wrote:

> Hi,
>
> From: "Nathaniel Talbott" <ntalbott@rolemodelsoft.com>
>
> > One off topic question is, why isn't alias_method() named copy_method?
Seems
> > clearer about what it actually does.
>
> Because what alias actually does is making an alias :-)
>
> As far as alias is concerned, you can view method definition
> just like assigning proc objects to variables.
>
>   foo = proc { Dave() }     # def foo; Dave(); end
>   bar = foo                 # alias bar foo
>   foo = proc { Fred() }     # def foo; Fred(); end
>   bar[]                     # => Dave
>
> I guess what you had in mind was something like below.
>
>   foo[0] = { Dave() }
>   bar = foo                 # alias bar foo
>   foo[0] = { Fred() }
>   bar[0][]                  # => Fred
>
> You see the alias part works in the same way, but end result is
> diffetent.

Ah, I see. If we think of the class as just holding a reference to a method
instance, alias_method is just adding a new method reference to the class
and pointing it at the same method instance as the original reference.
Changing what the original reference points at obviously doesn't change what
the new reference points at. Now that I understand that, the current
behavior makes sense, but it certainly doesn't seem obvious at first blush.
Dave & Andy, do you think it would be valuable to clarify this in the book?

It still seems that, whatever it's doing underneath, from a usage standpoint
the current alias_method() is copying the method, since I don't think
there's a way to change the original method such that it affects the
'alias'. Is this true?


> BTW, one common apprication of alias is inserting extra code
> into existing method.
>
> Something like....
>
>   class String
>     alias __to_s__ to_s
>
>     def to_s
>       "``" + __to_s__ + "''"
>     end
>   end
>
> This will end up infinite recursion if redefining to_s will
> change __to_s__ as well.

I certainly don't mean to imply that this behavior is bad; rather, it seems
extremely useful. I'm just wondering about the naming, and whether, as Jim
suggested, it's useful to have a way to create a 'true' alias. I'm certainly
happy with just overriding method_missing() when I want to do these types of
things, but it might be valuable to have a builtin facility for it, too.


Nathaniel

<:((><
+ - -						+ - -
| RoleModel Software, Inc. &		| EQUIP VI
| The XP Software Studio(TM)		|