Ok, following the recent "cache key" discussion, I would like to explain how
I believe Open Source XP works best.
I coded jboss1.0 in 2 iterations, with a clean restart in the middle based
on rickard Proxy and RemoteSerialized ideas. jboss1.0 worked, and worked
fast. We never got it to production since we moved all efforts to jboss2.0.
Rickard did a fantastic job of "factoring" all the ideas we had investigated
in jboss1.0 in a solid and modular architecture, something that was missing.
He then added some "out there" stuff such as the pervasive plugin
architecture, the interceptor design to flow naturally in container calls
and the "from the ground up" JMX architecture, the JMX stuff is truly
"ahead" of our time and the base for JSR77. These are truly revolutionnary
designs. It was by no means finished and beyond NYI some stuff needed
straightforward refactoring and even redesign (e.g. Jaws EJB calls).
In clear, ALL design is ALWAYS iterative and in 3-4 iterations of PEER
REVIEW we reach a kick ass level that proprietary stuff doesn't really want.
I think we have a great group of peers and the basic assumption needs to be
"trust". I have seen rickard "naked" he has seen me "naked", in the code
sense of course. And when we see each other naked, instead of saying "here
PooPoo" :)) we should say... let me fix this... you did a good first pass,
how about this second pass ??? and when I am done with it can you take a
third pass :)))
Point in case, the Cache Key stuff... Frankly and I think we are beyond that
point, the cache design was a bit "bare" and plain broken in 90% cases. NO
SWEAT, I know how to fix it... so I XP it last week. Is my solution
finished? no! Can I do better by refactoring the FastKey in CacheKey and
moving it all to OBject, as proposed in DaSolution and some of your points?
yes! Andreas was here Saturday and I gave him the little excercise of the
"proper" factorization, which he did in a cool XP way...
So let's continue working just like we did, but change our mindsets a tiny
bit (including me). We have a unique group of PEERS, an outstanding team of
world experts, and as an open source Team (some of it on Telkel staff) we
are putting forth something I feel is worth gold and we should collectively
work on.
PLgC
marc
________________________
Marc Fleury, Ph.D.
Chief Technology Officer
Telkel, Inc.
________________________
