From: Mark Hubbart <discord@...>
Date: 2004-04-20T09:48:49+09:00
Subject: Re: proposal: call_up() for use in redefined methods


On Apr 19, 2004, at 4:04 PM, Hal Fulton wrote:

> Mark Hubbart wrote:
>
> [snip]
>
>> Comments? Questions? Flames? :)
>
> I think this is essentially the same as my suggestion in
> http://ruby-talk.org/38136

Damn! you're right - and I like "prior" better, too :) I should've 
googled it a little better, might have turned that up.

> The answer then was basically: Something like this will come
> later.
>
> In fact, I believe that Matz's plans for pre/post/wrap would make
> this sort of thing easy (and far more generalized).

hmmm... from what I've read, it appears the hooks are just methods, and 
they can be redefined. So if you re-open a class and define foo:wrap, 
and foo was already wrapped, it looks like the new wrapper will replace 
the old one, rather than wrapping the old wrapper.

The beauty of the "prior" method is that you can add support for other 
object types easily, by redefining the method. So if you are writing 
libraries for HyperNumerics and PsuedoNumerics, you just go:

   # file hypernumeric.rb
   module Math
     redefine :sqrt
     def sqrt(obj)
       if obj.kind_of? HyperNumeric
         # handle hypernumeric numbers
       else prior(obj)
       end
     end
   end

   # file psuedonumeric.rb
   module Math
     redefine :sqrt
     def sqrt(obj)
       if obj.kind_of? PsuedoNumeric
         # handle psuedonumeric numbers
       else prior(obj)
       end
     end
   end

no method entanglement, no namespace clutter... And it doesn't matter 
which one gets defined first, since it just grabs the argument if it 
can handle it; otherwise it passes it on.

> Unless I've misunderstood.

Or me :) I might easily have misunderstood how the pre/post/wrap thing 
will work... I'm sorta unsure of it, the more I think about it...

--Mark