On Jan 6, 2009, at 10:44, WT wrote:
> It just occurred to me, though, that the KVO mechanism *could* do
> something special for dependent keys, namely, pass the original
> change dictionary on to the observer of the dependent key. After
> all, it does know that it's dealing with a dependent key. As it is,
> once I get the notification for the dependent key, all information
> about the original change is lost.
It's an interesting concept, but creating additional values would
increase the observation overhead. It's also not obvious that
coordinating (and coalescing?) independent key changes would give
consistent/useful results in the change dictionary.
The beauty of KVO is its, um, elegant minimalism.
