Although I like the idea of a size_t type that can later be redefined in a hypothetical future 128-bit situation, I’m afraid that that still will only be suitable for a subset of the uses of size_t. People will always make assumptions about the size of size_t, so I’m not so sure whether we should introduce yet another type that has a basically unknown size. After all, if you start to know the size of size_t, it’ll loose its ability to be scaled up on future hardware. It seems out of place in this proposal, given that the rest is basically going from unknown/flexible types like int and long to rigid types like int64_t.
Personally I’m a fan of int64_t, uint32_t, etc. as it takes away any doubt about the size of variables.
Als +1 on gradually moving from short/int to bool when it’s about boolean values.