Pour me rappeler, je m'essai aux résumés.

Prefer NRVO to RVO

* RVO = return-value optimization

MyData myFunction() {
    return MyData(); // Create and return unnamed obj
}
MyData abc = myFunction();

With RVO, the C++ standard allows the compiler to skip the creation of the temporary, treating both object instances—the one inside the function and the one assigned to the variable outside the function—as the same. This usually goes under the name of copy elision. But what is elided here is the temporary and the copy.

• NRVO = named return-value optimization

MyData myFunction() {
  MyData result;           // Declare return val in ONE place
  if (doing_something) {
    return result;       // Return same val everywhere
  }
  // Doing something else
  return result;           // Return same val everywhere
} 
MyData abc = myFunction();

Named Return Value Optimization is similar but it allows the compiler to eliminate not just rvalues (temporaries), but lvalues (local variables), too, under certain conditions.

But many compilers are actualy failing to apply NRVO with the following code:

 MyData myFunction() {
    if (doing_something)
        return MyData();     // RVO expected

    MyData result;

    // ...
    return result;           // NRVO expected
} 
MyData abc = myFunction();

So previous example has to be preferred.

Return parameters by value whenever possible

Surprisingly, don’t use "out-parameters" but prefer "return-by-value", even if it implies creating a small struct in order to return multiple values. I will check that, because it seems strange to me.

“out” parameter pointers force a modern compiler to avoid certain optimisations when calling non-inlined functions.

Prefer:

struct fractional_parts {
    int numerator;
    int denominator;
};

fractional_parts convertToFraction(double val) {
    int numerator = /*calculation */ ;
    int denominator = /*calculation */ ;
    return {numerator, denominator}; // C++11 braced initialisation -> RVO
}
auto parts = convertToFraction(val);
use(parts.nominator);
use(parts.denominator);

than:

void convertToFraction(double val, int &numerator, int &denominator) {
    numerator = /*calculation */ ;
    denominator = /*calculation */ ;
}

int numerator, denominator;
convertToFraction(val, numerator, denominator); // or was it "denominator, nominator"?
use(numerator);
use(denominator);    

Cache member-variables and reference parameters

Prefer:

template <class T> 
complex<T> &complex<T>;::operator*=(const complex<T> &a) {
   T a_real = a.real, a_imag = a.imag;
   T t_real =   real, t_imag =   imag; // t == this
   real = t_real * a_real – t_imag * a_imag;
   imag = t_real * a_imag + t_imag * a_real;
   return *this;
}

than:

template <class T> 
complex<T> &complex<T>;::operator*=(const complex<T> &a) {
   real = real * a.real – imag * a.imag;
   imag = real * a.imag + imag * a.real;
   return *this;
}

It seems the example is too simple to really understand the benefits. I do not want to make my code more complex for such a simple case... I do not like this advice.

Organise intelligently member variables

Due to the CPU caches mechanisms (blocks of 64-byte), it is propose to organise member variables in a class as follow:

• Move the most-frequently-used member-variables first
• Move the least-frequently-used member-variables last
• If variables are often used together, group them near each other
• Try to reference variables in your functions in the order they’re declared

Why not...

TBR