I see you & Berin's point Stephen, and I see Pete's. Both are valid. I think a third strategy might be a proper compromise, and allow us to move forward.
+1
there are good arguments either way. The problem is in settling on whether an abstraction is leaky or not. It is very *hard* to create a completely non-leaky abstraction; in fact it is probably impossible in a totally generic system.
Thus the problem we face is a choice in what can reasonably be considered general in avalon's solution space.
Look at any current or feature system that attempts to deal with stuff like this, and you will find conceptual compromise throughout. The questions then are
- are we willing to invest the time and energy neccessary to reach compromise?
- how are we going to go about finding that compromise?
- what do we do on failure to reach it?
Clearly, we tackled the problem the wrong way back in september, given the resulting 4 or 5 competing proposals which are distributed across cvses and projects, the flamewars, etc etc. The idea I guess a few of us have is that because of that failure, we're going to keep things container-specific. The route of least resistance, I guess.
Also, looking at the massive energy it cost to deal with our failure back then, is making many of us wary (me at least) from dealing with this stuff now, when we have lots of other things to work on as well.
My opinion is that the changes to fortress to incorporate the meta-stuff features should be rolled back or done on a post-1.0 branch, the discussion postponed, releases finished, and general quality of existing stuff improved (we really need unit tests!) before tackling this. That said, I won't be in the way. I'm just gonna take on a passive role here while you guys battle the meta once more. I don't feel like doing this right now :D
- LSD
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
