I like the thought of coroutines, but it's definitely been especially slow for the ecosystem to adapt to asyncio so far. (And for good reason. It has new syntax that doesn't play well with non-asyncio). Adding even more syntax for the same things may slow down adoption more? Admittedly, it's much better syntax.

The optional typing is a tricky subject. I'll have to wait until there are tools actually using it to see. One of my biggest hesitances with it is that until runtime checking or really good static analysis exists, annotating things incorrectly can be too easy. I am cautiously optimistic though. The theory is fantastic, just have to wait for the follow through. The best way to sort-of statically type things right now(imo) is through namedtuples (which I make use of heavily for lightweight, strict datatypes)

Formatting for bytes and bytearray is actually one I run into a ton trying to port py2 code to py3. Pretty excited about that.