Initialization of local variables: are these methods different semantically?

Question

In C++, is there any semantic difference between the following 3 methods of initialization?

T t;
T t = T();
auto t = T();

I'm interested in differences w.r.t. copy constructor, destructor and assignment operator behavior.

Answer 1

They are not equivalent. The first one leaves t uninitialized if it's a POD type, whereas the latter two will value-initialize the object no matter what. Example:

#include <iostream>

using namespace std;

int main()
{
    int a = int();

    cout << a << endl;
    return 0;
}

results in:

$ clang++ -O2 -o init init.cpp
$ ./init
0

whereas this:

#include <iostream>

using namespace std;

int main()
{
    int a;

    cout << a << endl;
    return 0;
}

will output some garbage (or crash, or make demons fly out of your nose) since it has undefined behavior arising out of the uninitialized object:

$ clang++ -O2 -o init init.cpp
$ ./init
1348959264

---

As to the question of copy constructors and assignment operators: the second and third snippets may invoke one of them (or may not, thanks to copy elision), so either of them (or both) need to be available.