From: "juan pedro meriño" <juapdiaz@...>
Date: 2007-06-22T05:23:05+09:00
Subject: Re: Modular classes and avoiding extension clashes

------=_Part_153073_19715328.1182457387201
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

I have ruby in my linux, but i don't know how do start to ruby for web,
please how i make it?

2007/6/21, Trans <transfire@gmail.com>:
>
> I refer back to:
>
>   Ruby-talk: 240334
>
>
>
> http://groups.google.com/group/ruby-talk-google/browse_thread/thread/9dbe=
8cd0898227b5/409301042321f6fc?lnk=3Dgst&q=3Dopen+class+module+trans&rnum=3D=
2&hl=3Den#409301042321f6fc
>
> I've thought of another potential benefit of this approach. While I
> had believed it would be only reasonable to have a single anonymous
> module per class, there may be greater benefit in multiple modules. B/
> c, it may make possible the avoidance of extension clashes. Here's an
> example:
>
>   # friend1.rb
>   class String
>     def exclaim; self + '!'; end
>     def scream; exclaim + '!!'; end
>   end
>   ...
>
>   # friend2.rb
>   class String
>     def exclaim; puts self + '!'; end
>     def scream; exclaim; exclaim; end
>   end
>   ...
>
>   # elsewhere.rb
>   require 'firend1'
>   require 'firend2'
>
> Clearly there's a conflict. How can I use friend1#scream but also
> friend2#scream? One's first thought could well be to alias the method
> so it's under a different name:
>
>   # elsewhere.rb
>   require 'firend1'
>
>   class String
>     alias :friend1scream, :scream
>   end
>
>   require 'friend2'
>
>   "hey".firend1scream
>
> However, #exclaim is called by #scream so it will still not work as
> expected.
>
> So getting back to the original suggestion (of the previous thread),
> having open extensions contained in hidden modules... what if method
> lookup followed a route of first checking within the same module, then
> continuing upward before dropping back to ancestors? In other words if
> we had a class, C < M1 < M2 and a method was called on it that is
> defined in M2, then any method the M2 method calls would first be
> looked for in M2, then M1, before falling back to Object and Kernel.
> This would resolve the problem with the above example.
>
> It seems to me this is not necessarily unexpected behavior either. If
> I create a method A that depends on a another method B, clearly
> overriding B is dangerous business --it's something not generally
> done. Usually it's the "top most" methods --the ones that depend on
> others, that we override. That's not to say there aren't cases to do
> so, but in that case a 'redef' keyword could be used explicitly.
>
> T.
>
>
>


--=20
Juan Pedro Meri=F1o Diaz
Ing de Sistema CURN

------=_Part_153073_19715328.1182457387201--