From: "Chr. Rippel" <chr_news@...>
Date: 2002-01-12T04:04:38+09:00
Subject: Re: class name reported differently in different contexts

"ts" <decoux@moulon.inra.fr> wrote in
....
>  Don't forget : if I've a classpath, I can retrieve the constant actually.
>  If I have only a classid, I can't retrieve the constant.

So what you are saying is that you can do something like
------
begin
  class Klass
    def self.inherited(sub_klass)
      const_get sub_klass.inspect.sub(/^(Klass::)/, '').intern
       # before Nobu's patch it's even simpler - i.e.
       # const_get sub_klass.intern
       # this would be an argument of neither setting classid nor
       # classpath
    end
    class Sub < Klass;end
  end
rescue NameError => mes
  puts mes # => uninitialized constant Sub at Klass
end
------

>  With this patch, classid is useless and even for "native" classes it can
>  exist only classpath.

Yes, following the logic of this patch you might as well merge classid
and classpath into one variable (as far as I understand it's implications).

Playing the devils advocate there is actually some reason for setting
classpath (and not classid) in the second form of class creation

------
module Outer
    klass = Class.new
    # if the this call is succesful the classpath of klass
    # is know and could already be set.
end
------

It also seems plausible that there exist a situation where you would
be in the position of rescuing an exception imminating from an
#inherited call and it seems that the only way of correctly defining
the corresponding class constant is that classid is already set before
the call to #inherited - i.e. the previous behavior.


/Christoph