[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ruby-talk
Subject:    Re: [ANN] Dissident 0.1, a Ruby dependency injection container
From:       Jacob Fugal <lukfugl () gmail ! com>
Date:       2005-09-22 14:37:39
Message-ID: 5075676705092207373a40bdf2 () mail ! gmail ! com
[Download RAW message or body]

On 9/22/05, Christian Neukirchen <chneukirchen@gmail.com> wrote:
> Piergiuliano Bossi <p_bossi_AGAINST_SPAM@tiscali.it> writes:
> > Ehm, to be precise I'm not advocating the use of an IoC container; on
> > the other side, the principle of dependency injection is an important
> > tool to put in your box.
>
> Just curious, but you could explain why you do not advocate using IoC
> containers?

In my understanding, Giuliano wasn't advocating against using IoC
frameworks[1] (or more specifically, DI frameworks[2]). He simply
disclaimed himself from advocating a framework or frameworks in
general. He advocates DI as a principle/tool but leaves it at that.

Of course, he may indeed have been advocating against using DI
frameworks, in which case I'll let him go ahead and defend his
position. :)

Jacob Fugal

[1] I replaced discussion of "containers" with "frameworks". A
container is an element of one approach to dependency injection. It's
not a necessary element, and even when it is used it may be named
differently (e.g. Coplands "registry"). Also, ad-hoc DI without a
framework may use a form of container. What's being discussed -- ie.
entities such as Copland, Needle, Dissident, Matz-DI and even Java's
PicoContainer (despite the word "Container" in the name) -- are
frameworks.

[2] The confusion about Inversion of Control (IoC) versus Dependency
Injection (DI) is significant. I'd suggest avoiding the term IoC in
favor of DI unless you are specifically talking about an IoC principle
which is independent of DI.


[prev in list] [next in list] [prev in thread] [next in thread]