Consider the following situation:
#include <algorithm>
#include <vector>
class T {/*...*/};
bool predicateA(const T& t) {/*...*/}
bool predicateB(T& t) {/*...*/}
bool evaluate(std::vector<T>& vA, std::vector<T>& vB)
{
return std::ranges::count_if(vA, predicateA) + std::ranges::count_if(vB, predicateB);
}
If std::ranges::count_if(vA, predicateA) returns a positive value, then evaluate will return true no matter what the other call to count_if returns. Would the compiler be required to evaluate the second call to count_if anyway? Notice that in this particular situation eliding that call might change the behavior of the program, since predicateB is allowed to have side effects.
No, certainly not. The only short-circuit evaluation is that defined by the language like for && and ||.
0