Nikolai Weibull <mailing-lists.ruby-talk / rawuncut.elitemail.org> writes:

> OK.  So Ruby has enough dependency-injection/inversion-of-control
> libraries as it is, but mine is quite different so I figured I¡Çd post
> it.  It¡Çs about as simple as dependency-injection can possibly become
> and there¡Çs no support for lifecycles, interception, or other framework
> stuff.  Nor is there any support for parents, children, or other
> relatives, so keep it to yourself.

Interesting idea, but you are only pushing constants around?
Can't we do that easier?

> class Scanner
>   def initialize(io)
>     raise Error, "can only work with open io-ports" if io.closed?
>     @io = io
>   end
> end

> class Error < ScriptError; end # Or something more creative, perhaps?
  Scanner::Error = Error

> class TC_Driver < Test::Unit::TestCase
>   ãë¥ç
> end

> So the idea is that under normal circumstances, ¡Æneeds¡Ç is just a
> circumvention for not using ¡Ærequire¡Ç directly, but for trying times the
> dependency can be changed to something more suitable.

I really fail to see the advantage of using this over manual require
and dependency setting. :-)

I personally would expect a DI framework to at least provide a
singleton instantiation scheme, else it wouldn't make any sense to me
at all.

Einstein once said "Everything should be made as simple as possible,
but not simpler".  I think anything more simple than Matz-DI is too
simple.
> http://onestepback.org/articles/depinj/matz/index.html

>         nikolai
-- 
Christian Neukirchen  <chneukirchen / gmail.com>  http://chneukirchen.org