From: TRANS <transfire@...>
Date: 2005-09-27T06:40:23+09:00
Subject: Re: Require Namepaces and RubyGems' effect on LoadPath problem

On 9/26/05, Austin Ziegler <halostatue@gmail.com> wrote:

> Two things. 1. I don't see any particular reason for this. It's neat,
> but I don't see why it's necessary. 2. If something is loaded from a
> particular "namespaced" path, that path should remain on the load
> list, because not everything in a path may actually be loaded with a
> single require.

It circumvents the need for a #require_local. In doing so it reduces
"dependencies" in lib files. What I mean by that is I don't have to
specify absolute paths in internal lib files. For instance in Nano. 
'nano/integer/even.rb' was:

  require 'nano/integer/odd'
  require 'nano/integer/multiple'

  class Integer
    def even?
      ...

But now with this is can simply start:

  require 'integer/odd'
  require 'integer/multiple'

Having added 'nano' => [ 'nano' ] namespace.

As for #2, they can't remain on the load path, b/c it would fill it up
with local locations. You just reuse the namspace it you have another
one to load:

in my program:

  require 'nano' # adds the namespace
  require 'nano:integer/odd'
  require 'nano:integer/factorial'

Finally, it also allows one to organize ones lib directory
independently of how they are required by the end-user. For instance:

  $LOAD_SPACE['somelib'] = %w{
     somelib/onedir
     somelib/onedir/subdir
     somelib/twodir
  }

T.