Which Casting Method Works Best

Question

void Generic::IntToS16 (S16 & Out_, int & In_){
    unsigned char * bytes_in = reinterpret_cast<unsigned char*> (&In_);
    unsigned char * bytes_out = reinterpret_cast<unsigned char*> (&Out_); 
    bytes_out[1] = bytes_in[1];
    bytes_out[0] = bytes_in[0];

    bytes_out[1] |= bytes_in[sizeof(int)-1] & 1<<7;
}

---OR---

short Generic::IntToS16 (int In_){
    short ret = static_cast<short>(In_);
    return ret;
}

Answer 1

The first method is not a cast, it's an architecture-dependent operation that will produce invalid results on litte-endian architectures. In other words, it is not portable.

The second cast, however, is portable, because it lets the compiler generate the correct sequence of operations.

Answer 2

OR

short Generic::IntToS16 (int In_){
    return In_;
}

Note: short is not necessarily 16 bits long (it is at least 16 bits but could be longer).

You can of course use: int16_t

Also note casting from a large integer type to a smaller may result in implementation defined values:

If the destination type is signed, the value is unchanged if it can be represented in the destination type (and bit-field width); otherwise, the value is implementation-defined.