The final solution to the dependency injection problem might just have been discovered by Brian McCallister. After many trials and errors to correctly dissociate configuration from implementation, a legacy of numerous IOC types, Brian came up with a mechanism that weaves in at runtime and requires no XML configuration at all.

Imagine fixing a bug via...

update
    methods
set
    method_body='return new Integer(42);'
where
    package = 'org.example.foo' and
    name = 'getDayOfWeek';

More details can be found here:
Inverting the Inversion of Control

note: take this entry with a grain of salt ;) pre-friday humor

Hopefully indentations in the sample where made to a foreseen COBOL implementation.

...as soon as it has a UML/MDA Modeling plugin for Eclipse.

And please, be sure to throw only RuntimeExceptions ;-)

...as soon as it has a UML/MDA Modeling plugin for Eclipse.And please, be sure to throw only RuntimeExceptions ;-)

Yes, IoC 2D is the best type, but XML is a cool programming language too.

I was about to let it go, but I can't, just can't.

[...] but XML is a cool programming language too.

w3.org servers glitched for a sec, there...

I'd very much like to know what do you compile your XML with. And please DON'T clarify saying "XML is a declarative programming language, c'mon; a man came out of my coffee cup and told me so!".

I was about to let it go, but I can't, just can't.

[...] but XML is a cool programming language too.

w3.org servers glitched for a sec, there...I'd very much like to know what do you compile your XML with. And please DON'T clarify saying "XML is a declarative programming language, c'mon; a man came out of my coffee cup and told me so!".

Actually if you are developing a UI with Lazslo or with Macromedia's Flex solution you actually do "program" with XML, and you do actually have a compiler that compiles it.

Dave Wolf
Cynergy Systems

Actually if you are developing a UI with Lazslo or with Macromedia's Flex solution you actually do "program" with XML, and you do actually have a compiler that compiles it.Dave WolfCynergy Systems

Nightmares come true.

I guess you could consider every piece of note you scrap in a postick and stick to your monitor is "programming", but let's get real...
XML by itself is no programming language. If and only if you augment it with semantics it COULD be considered a programming language, why not? But certainly not XML alone; at least that's my opinion.

Regards,
Martin

I was about to let it go, but I can't, just can't.

[...] but XML is a cool programming language too.

w3.org servers glitched for a sec, there...I'd very much like to know what do you compile your XML with. And please DON'T clarify saying "XML is a declarative programming language, c'mon; a man came out of my coffee cup and told me so!".

Sorry, if this kind of humor hurts.

Sorry, I didn't get what you meant. But I apologize if my remark was a bit acid... I was just being a smartass. ;)

Regards,
Martin

Yes, IoC 2D is the best type, but XML is a cool programming language too.

Huh ?

*loud pop* all my code has just turned in on itself and disappeared. Nice one brian, a new technique and code reduction - beautiful ....

Is TSS entering comedy business under the new managerment?

So what? instead of using XML config, you use a database table to dynamically change the method string. Why not Groovy?

Perhaps its things like

(in-package 'foo)
(compile (defun get-day-of-week () '42))

I can see a great potential for applying this technique in the next generation of automated voting machines. Unfortunately those backwards parts of the world where they still use paper ballots will miss out (sob!!).

Some of us may be scared about the performance implications of a virtual virtual-table lookup. Happily, Java Five's -javaagent flag comes to the rescue, by allowing the bytecodes of the allegedly buggy classes be redefined dynamically. Which starts to look shockingly similar to a microboot.

-Patrick

--
Patrick Linskey
Kodo JDO
http://solarmetric.com

Why reinvent Smalltalk when there is couple of them already?

Recall one of the advantage of IoC is to have the implementation hidden from the user of the class (or service?) by only showing what it needs thru either constructor or a set method. This makes use of interface programming. One concern comes out of my mind is that whether we will then "new" the implementation or the interface (which is prohibited by compiler i think). I'm not sure if some AOP weaver could help in this area to make it compilable.

My 2 penny