The easiest way of implementing a string move semantics is simply to steal the internal byte representation pointer of the other element, and put an empty one in its place.
This is just intuition, but I was curious to check it myself. Stupid snippet to check this out:
#include <iostream>
#include <string>
int main()
{
std::string s1 = "hello";
std::cerr << "1 " << s1 << " -> " << (void*)s1.c_str() << std::endl;
std::string s2 = std::move(s1);
std::cerr << "1 " << s1 << " -> " << (void*)s1.c_str() << std::endl;
std::cerr << "2 " << s2 << " -> " << (void*)s2.c_str() << std::endl;
}
The surprising result:
1 hello -> 0x7ffce8e6a790
1 -> 0x7ffce8e6a790
2 hello -> 0x7ffce8e6a770
Not exactly what I thought… a new string really. Could it be because
"hello" is so short?
#include <iostream>
#include <string>
int main()
{
std::string s1 = "hello world of people long long";
std::cerr << "1 " << s1 << " -> " << (void*)s1.c_str() << std::endl;
std::string s2 = std::move(s1);
std::cerr << "1 " << s1 << " -> " << (void*)s1.c_str() << std::endl;
std::cerr << "2 " << s2 << " -> " << (void*)s2.c_str() << std::endl;
}
How about this long string now?
1 hello world of people long long -> 0xebec20
1 -> 0x7ffd740008e0
2 hello world of people long long -> 0xebec20
Interesting :)