Does that also mean that it will have better support for timeouts? I'm not interested in socket timeouts, more in a 'finish this in N seconds or just give up' thing.

For that reason I am using pycurl now for some stuff where I had previously used the requests library. Which has a really really horrible API :-/