avalon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Courcoux <pe...@courcoux.biz>
Subject Re: [Proposal] Last ditch effort for one Java codebase
Date Tue, 05 Aug 2003 19:44:34 GMT
0.00002 cents worth from a lurker!

Random thoughts follow!

AF, Phoenix and Fortress are released. You're stuck with them!

Please don't compound the misery by releasing anything else from Avalon
until problems are resolved.

Avalon is 'accident prone'. Events of the past year have shown that. A
reusable framework is advanced stuff. I don't understand all the
concepts (I'm no expert and wouldn't expect to) but from what I read
others find the learning curve steep as well. 

Highly competent developers such as those developing at Avalon,
including Peter Donald and Stephen McConnell probably have a vision of
the future and an understanding of the implications of what they are
doing far in excess of many others. I would not expect them to give way
easily when they believe in what they are saying. That does not mean
that they are always right! I also suspect that they are developing
Avalon containers to use not for any research or thesis but to make
their own real commercial work easier.

Getting such envisioned people to agree, under pressure, is never going
to be easy. 

Peter and Stephen have both fought hard to stay within the Avalon fold.
Phoenix appears to be almost at a standstill, which is a shame and
Stephen is fighting hard for Merlin, which it has been suggested may
well have to move out of Avalon. Avalon must have something. Lets try
and build on it.

Would it be possible to produce some container specifications which
would allow the visions of all the Avalon developers to be met by the
proposed component based container, even to plugging in differing
standards of meta information, for example?

Would there then be a way forward for componentisation of existing
containers providing backwards compatibility for Phoenix, Fortress and
Merlin within differing configurations of the one container?

If so, the roadmap for 'one container' looks good to me and it is up to
the various developers to produce the components they need for their own
container. Hopefully, there would be much reuse and many components
written for the merlin-one-container could be used for the
phoenix-one-container and the fortress-one-container etc.

If that is feasible, then hopefully clashes can be avoided and all can
be accommodated. 

Clearly, such a move may further delay release of Merlin, but I would
have to question the wisdom of making a release of Merlin under Avalon
unless the underlying causes of the last 12 months 'accidents' are
resolved. Papering over the cracks won't stop them getting bigger. 

Hopefully, the value of 'Apache Avalon', would encourage developers to
develop container components meeting the Avalon Container
Specifications. If Stephen needed to release Merlin outside of Avalon,
hopefully, many developers would help with creating components to
effectively bring it back in again as a series of components.

It would also be good to create an environment in which Peter would feel
he would like to contribute.

I think that much of the above is what is intended by the one-container

The trick is to make the container specs. able to handle, at least
components to meet Phoenix, Fortress and Merlin requirements. 

As I said above, Avalon seems 'accident prone'. I suggest that asking
anyone to back down may not be the right thing to do. Maybe the only
approach is to be sufficiently flexible to be able to accommodate
alternative visions. Is the one-container proposal able to accommodate
the differing visions? 

Avalon users want the benefits of IoC, SoC, and the pluggability of
alternative implementations. It looks like this is what the container
developers need.

I must be missing something!

Maybe it's only worth 0.000000000002 cents.


On Tue, 2003-08-05 at 18:32, Berin Loritsch wrote:
> Leo Simons wrote:
> > Berin Loritsch wrote:
> > 
> >> At this point I think it would be futile to start up a new effort and
> >> port features to a container framework.
> > 
> > 
> > can you summarise why? (in a different thread, perhaps)
> It seems there isn't enough interest in a new effort.  Everytime it is
> brought up the folks who want a multi-container in Avalon environment
> keep shouting it down.  I don't think we have enough developers for all
> the containers we have now.  Something has to give, so I am not going
> to press my opinion on a fresh start.
> > 
> >> Does this sound like a decent plan of action?
> > 
> > 
> > I am not sure. It sounds like a reasonable compromise between all the 
> > other options we've talked about so far. Concern: I'm wondering whether 
> > this will generate the positive energy we need: working on new code is 
> > usually more fun (again, IMO) than working on existing code.
> > 
> I agree, but it is the only thing I can think of if we are going to salvage
> this community and get back to having fun.
Peter Courcoux <peter@courcoux.biz>

To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org

View raw message