I know, but projects like `greenbean` just further encourage the mindshare that async doesn't have to be the only way to do hi perf networking. Especially if normal OS threads can be made just as cheap as a stack of Futures without a special runtime like Go requires.
Also this setup might be cheap enough to do a thread per request and do a true actor model of one thread per actor. As I understand it Mummy and others still use a worker pool right? That was my impression at least. Nothing wrong with that, I just enjoy the actor model for that sorta stuff.
