Hello,

I'm having a look into that Spring stuff and therefore into dependency
injection, of course. Now I'm wondering what's the great catch about
DI. Yes, right, you get around import statements in your code that
reference technology-specific or vendor-specific classes. This is of
course very useful. What I actually already did myself without knowing
about DI was interface-based injection which is an obvious approach I
would say that everyone would find himself. I was skimming my Spring
book from beginning to end and back again and it seems to me that DI
is not much more than that. I mean build a XML parser that can resolve
references to other objects defined in XML and you are basically done
(objects referenced must be programmed against an interface, of
course, to be decoupled).

I thought that DI was something like this:

public class MyBean
{
   @PersistenceContext(name="myDB")
   EntityManager em = null;
}

Now when I create an instance of MyBean the variable em is already
set. I don't have to bother how to get a reference to that thing and
am therefore decoupled from the respective JPA provider. How does em
get set now? Seems to me that this must be done when the lookup from
the JNDI-Server (InitialContext) is made. But then I have a dependency
on the JNDI-Server, I'm not really technology independent here.

Maybe someone out there could help a poor one-eyed guy out with this
and explain this to me. Maybe I'm just thinking too far and thinks are
perfect the way they are ...

Thanks, Olli