Just to cut this thread short, I'm going to reject PEP 505, because ? is just too ugly to add to Python IMO. Sorry.
I commend Mark for his clean write-up, without being distracted, giving some good use cases. I also like that he focused on a minimal addition to the language and didn't get distracted by hyper-generalizations.
I also like that he left out f?(...) -- the use case is much weaker; usually it's the object whose method you're calling that might be None, as in title?.upper().
Some nits for the PEP:
- I don't think it ever gives the priority for the ?? operator. What would "a ?? b or c" mean?
- You don't explain why it's x ?? y but x ?= y. I would have expected either x ? y or x ??= y.
- You don't explain or show how far ?. reaches; I assume x?y.z is equivalent to None if x is None else x.y.z, so you don't have to write x?.y?.z just to handle x.y.z if x is None.