Continuum architecture management

[Wendy Smoak-3]

Since we're talking about making architectural changes, I wonder how
close we really are to the "now" diagram on
http://continuum.apache.org/ref/1.4.0-SNAPSHOT/architecture.html ?

I tried looking at Continuum with Structure101 late last year, but
didn't get very far because the packages are still split between
org.apache.maven.continuum and org.apache.continuum.  What I _could_
see tells me that we have a some dependencies between modules that
really shouldn't be allowed to talk to each other. :)

Recently I've been trying this again with SonarJ, which the nice folks
at Hello2morrow have provided a license for.  With this tool, you
define your own layers and  rules about which layers are allowed to
communicate, and then it shows you where the dependencies are.

I would really like to define the new architecture and use automated
tools to make sure we adhere to it (or at least know when we aren't).

Thoughts on this in general?  Any other tools we should evaluate?

--
Wendy

[Emmanuel Venisse-2]

It is a good thing to analyze the architecture and if possible to automated
it.

other tools are:
- Latttix but it seems to be very poor
- xdepend, this tool seems to be very good but I'm not sure it generate
architecture diagram

We can get a free license for each of them.
Emmanuel

On Thu, Jun 25, 2009 at 5:14 PM, Wendy Smoak <wsmoak@...> wrote:

> Since we're talking about making architectural changes, I wonder how
> close we really are to the "now" diagram on
> http://continuum.apache.org/ref/1.4.0-SNAPSHOT/architecture.html ?
>
> I tried looking at Continuum with Structure101 late last year, but
> didn't get very far because the packages are still split between
> org.apache.maven.continuum and org.apache.continuum.  What I _could_
> see tells me that we have a some dependencies between modules that
> really shouldn't be allowed to talk to each other. :)
>
> Recently I've been trying this again with SonarJ, which the nice folks
> at Hello2morrow have provided a license for.  With this tool, you
> define your own layers and  rules about which layers are allowed to
> communicate, and then it shows you where the dependencies are.
>
> I would really like to define the new architecture and use automated
> tools to make sure we adhere to it (or at least know when we aren't).
>
> Thoughts on this in general?  Any other tools we should evaluate?
>
> --
> Wendy
>

[neerajsangal]

As somebody from Lattix, we will be happy to provide you with a license. Lattix will allow you to analyze your system and will also let you make changes to it to see the impact of those changes. Emmanuel, I appreciate your perspective; I will be happy to do a demonstration for you over the web and hear what you don't like about Lattix.

Neeraj Sangal

Emmanuel Venisse-2 wrote:

It is a good thing to analyze the architecture and if possible to automated
it.

other tools are:
- Latttix but it seems to be very poor
- xdepend, this tool seems to be very good but I'm not sure it generate
architecture diagram

We can get a free license for each of them.
Emmanuel

On Thu, Jun 25, 2009 at 5:14 PM, Wendy Smoak <[e-mail]> wrote:

> Since we're talking about making architectural changes, I wonder how
> close we really are to the "now" diagram on
> http://continuum.apache.org/ref/1.4.0-SNAPSHOT/architecture.html ?
>
> I tried looking at Continuum with Structure101 late last year, but
> didn't get very far because the packages are still split between
> org.apache.maven.continuum and org.apache.continuum.  What I _could_
> see tells me that we have a some dependencies between modules that
> really shouldn't be allowed to talk to each other. :)
>
> Recently I've been trying this again with SonarJ, which the nice folks
> at Hello2morrow have provided a license for.  With this tool, you
> define your own layers and  rules about which layers are allowed to
> communicate, and then it shows you where the dependencies are.
>
> I would really like to define the new architecture and use automated
> tools to make sure we adhere to it (or at least know when we aren't).
>
> Thoughts on this in general?  Any other tools we should evaluate?
>
> --
> Wendy
>