hm, so I tried this on the playground and it seems like Rust iterators also scale super-linearly in debug mode (which is expected, because every adaptor creates a new local that contains the previous one plus something extra), but this seems to go away in release mode (as everything is inlined). Is the problem in C++ something different?
it's exactly the same behaviour in C++. but people are used to this behaviour in Rust while in C++ they're used to writing traditional for-loops which do not have such an overhead at -O0 over -O3
