From: Yossef Mendelssohn <ymendel@...>
Date: 2008-05-06T00:10:18+09:00
Subject: Re: inherited behaviour.

On May 1, 3:37 pm, "David A. Black" <dbl...@rubypal.com> wrote:
> Hi --
>
>
>
> On Fri, 2 May 2008, Yossef Mendelssohn wrote:
> > On May 1, 11:32 am, Joel VanderWerf <vj...@path.berkeley.edu> wrote:
> >> Peter Morris wrote:
> >>> Folks, would there be any DOWNSIDE to the call to the parents INHERITED
> >>> method being deferred from the start to the end of the definition of the
> >>> subclass?
>
> >>> At the moment if you...
>
> >>> class A
> >>>  def self.inherited klass
> >>>  end
> >>> end
>
> >>> class B < A # this is point (1)
> >>>  def self.some_method
> >>>  end
> >>> end # this is point (2)
>
> >>> at point (1), the inherited call on A happens.
> >>> Now, what I want to know is.... would there be a problem if this was
> >>> deferred to point(2)
>
> >> Ruby classes are never closed, so there is no significant point (2),
> >> just an arbitrary snapshot in the continuing definition of class B.
>
> > There's only one inheritance point for a class, though. That the class
> > isn't closed doesn't even enter into it. It's not like you can change
> > the inheritance later on. Consider the following:
>
> >   class B < A
> >     # some stuff
> >   end
>
> >   class B < A  # reopening with explicit parent, fine
> >     # more stuff
> >   end
>
> >   class B  # reopening without parent, fine
> >     # more stuff
> >   end
>
> >   class B < C  # reopening with explicit parent, no good since the
> > parent is different
> >     # more stuff
> >   end
>
> > In this case, the "significant point (2)" would be the first 'end' in
> > this example, the one immediately following the "some stuff" comment.
>
> That seems very arbitrary, though, in the sense that the superclass's
> business with the subclass has no more to do with methods defined the
> first time the class is opened than with any others.
>
> David
>
> --
> Rails training from David A. Black and Ruby Power and Light:
>    INTRO TO RAILS         June 9-12            Berlin
>    ADVANCING WITH RAILS   June 16-19           Berlin
>    INTRO TO RAILS         June 24-27           London (Skills Matter)
> Seehttp://www.rubypal.comfor details and updates!

It is arbitrary, but it makes sense if the method is called
'inherited'. Really, I agree that it work the way it does now, when
the inheritance actually happens. If something has to happen at a
different point, it should be a different callback that does it.

--
-yossef