Bottom line regarding the CVS/Sandbox thing for me is that:

 1. We have no clue how this thing is going to look.

 2. We really don't know the requirements yet.

I think starting with a branch is at this point useless. I'd like
us to simply talk about what it is we want. We want pluggability,
etc. I think that's something everyone will agree on. Whether we
solve it as I *think* I can via Aspects/Appliances or some other 
way is still to be determined. I think I have solved two use cases,
but that's no guarantee for having solved them all.

For example, if configuration is one Aspect, who configures the
Aspects? How do Aspects interact with Appliances? Are Aspects
Appliances themselves?

There are millions of open issues surrounding this - I think we
should focus now on having a big think about this goal - pluggabilty -
and try to scope in in our heads before we sit down and start 
churning out code. We really don't need to code - we all know
all relevant language features in Java, so I'm sure we can code
anything we can come up with. Yes, we are that good. The trick
is to figure out just what we're going to code.

The goal was to have an architecture that was dead simple to 
explain. Therefore, any solution that can't be explained in an
email to the list sort-of disqualifies, in my opinion.

I'll let you know what I come up with. In a *short* email, I promise. 
I expect the rest of you to do the same. And remember - keep it 
simple and short. (That's the *hard* part.)

But please, nobody start coding up a storm in CVS now... It just 
becomes yet another one man show...

Enough preaching!

Aaron, you signed up for being Chair, so could you chair this 
discussion? We need someone to summarize and keep track of all
ideas, otherwise we'll just have a who-can-shout-loudest primordial
chaos. If you don't have the time, can I?

/LS


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to