Just to be clear, I am in favor of introducing DI frameworks in Hive where it makes sense. As Attila said, we don't want to get stuck with legacy code forever. When a concrete proposal comes up we can discuss benefits vs drawbacks.
Regarding stability I agree it is a pressing issue but Hive is an open source project and we certainly don't want to force volunteers to work on specific things or forbid them to work on others. Contributing to open source is supposed to be a fun and rewarding experience. I am sure many of the people in this list have stability as a primary goal so eventually we will get there. Best, Stamatis