Rupert Smith wrote:
Hi,
Paul Hickey <[EMAIL PROTECTED]>
Says that you can mail him to get a free for open source use license for
structure 101:
Just download from
http://www.headwaysoftware.com/downloads/structure101/index.php and shoot me
over your machine id (structure101/help/about/contact us).
The offer is straightforward. He would like us to add an acknowledgement,
similar to the one we already have for JProfiler, onto the site. If you try
it and like it, he would be happy to get some appreciative feedback for his
JOLT nomination, though there is absolutely no obligation for us to do so.
Also, at some point he might be able to make his repository of OS projects
that he has run 101 on available for us to upload to:
If you wish we can also provide you with the Web App and you can host your
own repository or if you want to make some suggestions as to the snapshots
you want on www.structure101.com and/or updated architecture diagrams just
let me know. Our plan is to give the different projects write access to
structure101.com but that ability is not yet there.
In particular, I note that the architecture that is on that site is probably
just one that was generated automatically, in a least-errors-fit way by the
tool? rather than what makes the most sense to us. The way to read that
diagram, is that is is a set of layers, where higher layers may depend on
lower ones, but not on ones at the same or higher levels. All the arrows you
see on it, are violations of this rule. Could at the very least, pull out
something high level that looks more like how we picture Qpid to be, that
is, common at the bottom, and client/server on top of it, and jms on top of
client, (tests/examples on top of everything), before adding back in some of
the detail.
I think it is worth a look if you haven't already.
I've found it useful in the past particularly for spotting bad
dependencies between layers in the architecture. Often when the
generated diagrams don't separate out naturally into the layers you
expect, its because the dependencies in the code aren't really
structured the way you think they should be.
