Hi Shane, > In AI we often can't even agree on this. Thus I see AGAGIs prprimarilya design > problem > rather than an implementation one. In Linux there is > a huge amount of boring implementation that needs to be done to get > all the drivers for hundreds of kinds of graphics hards and hard disk > controls etc. all coded and working. This is a very different problem.
Good point. The initial system will lock in some design assumptions, and people who join the project must buy into those decisions. But a good design can put a lot of decisions into modules that can be changed. I seem to recall Ben talking about major changes in Novamente modules. In my last visualization system, a major design goal was to put as many design decisions as possible in classes where they could be over-ridden. But you are right that some things are locked in, so enthusiastic developers are limited to those folks who like those locked-in decisions. Cheers, Bill ------- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]