Sort:  

I'm very skeptical on their rocksdb timeline. It looks like an estimate to just to write the code, without assuming adequate time for building a good test framework.

I've privately been advocating starting with SMT-Lite and rolling out more features later, so I'm happy with that change.