Grant,

I agree with your point on focusing the rearchitecture effort on a minimum amount of code that will give volunteer developers a solid foundation to do the rest. But I would put forward a slightly different definition of what that minimum is: How about initially implementing as little of the UI as possible and putting the bulk of your effort into the Domain and basic Storage layers?

This would give us a working, well-tested Chandler "engine" that could do CRUD operations on notes, tasks, and calendar items, plus sharing and a callback mechanism for periodic tasks and reminders.

Interested plugin developers would then have the option to write alternate Chandler UIs. For example, I'd be interested in exposing a DBus interface to Chandler, running it headless, and then using a collection of desktop widgets for viewing the calendar, quick entry, etc.