Prev: Objects Identification and Class diagram
Next: Mapping Class Diagram to coding
From: Corno on 2 May 2007 14:25
Hi all,

Some time ago (I think it was about 2 to 3 years) I read an interesting
article about Java metrics, but I cannot recall the name of the article and
I had no luck searching for it.
Does anybody know the article I'm referring to?

The article was about a method of analyzing the 'oo-ness'/coupling of a java
application. It came with a tool that calculated that level of oo-ness.
There were some examples of which (of course) only the source code of the
tool itself scored a 100% :) If I recall correctly, 2 others were Junit
(high score) and Eclipse (low score).

The article explained the opinion of the author that a package in Java
should either completely be an 'interface' (consisting of only interfaces
and abstract classes) or completely be an implementation of an interface
package without other packages depending on it.
Strict adherance to these rules introduced a problem during initialization
(the package with the main procedure -in an implementation package- will
have to construct classes in other implementation packages). The way to
solve this was a staged initialization procedure.
That's about as much as I remember of it. Anybody?

TIA,

Corno




From: Ed on 3 May 2007 15:53
On 2 Maj, 20:25, "Corno" <Corno(a)dds%FAKE%.nl> wrote:
> Hi all,
>
> Some time ago (I think it was about 2 to 3 years) I read an interesting
> article about Java metrics, but I cannot recall the name of the article and
> I had no luck searching for it.
> Does anybody know the article I'm referring to?
>
> The article was about a method of analyzing the 'oo-ness'/coupling of a java
> application. It came with a tool that calculated that level of oo-ness.
> There were some examples of which (of course) only the source code of the
> tool itself scored a 100% :) If I recall correctly, 2 others were Junit
> (high score) and Eclipse (low score).

Interesting: I didn't actually have analyses of those on the site, but
I'm in the process of doing it, now.

>
> The article explained the opinion of the author that a package in Java
> should either completely be an 'interface' (consisting of only interfaces
> and abstract classes) or completely be an implementation of an interface
> package without other packages depending on it.
> Strict adherance to these rules introduced a problem during initialization
> (the package with the main procedure -in an implementation package- will
> have to construct classes in other implementation packages). The way to
> solve this was a staged initialization procedure.
> That's about as much as I remember of it. Anybody?
>
> TIA,
>
> Corno

See below.

..ed

--

www.EdmundKirwan.com - Home of The Fractal Class Composition

From: Corno on 4 May 2007 04:32
"Ed" <iamfractal(a)hotmail.com> wrote in message
news:1178221996.445827.253150(a)n59g2000hsh.googlegroups.com...
> On 2 Maj, 20:25, "Corno" <Corno(a)dds%FAKE%.nl> wrote:
---
>> There were some examples of which (of course) only the source code of the
>> tool itself scored a 100% :) If I recall correctly, 2 others were Junit
>> (high score) and Eclipse (low score).
>
> Interesting: I didn't actually have analyses of those on the site, but
> I'm in the process of doing it, now.
>
---
>
> www.EdmundKirwan.com - Home of The Fractal Class Composition
>
Hi Edmund,

Thanks, this was exactly the article I was looking for. As I said;
interesting!

Corno



 | 
Pages: 1
Prev: Objects Identification and Class diagram
Next: Mapping Class Diagram to coding