Why only double?

When we now try to use sum0, we'll run into problems pretty soon:

double a[10];
float  b[10];
// ...
double sa = sum0(a, 10); // OK
float  sb = sum0(b, 10); // compile error

Unfortunately, sum0 works only with double, but not with float arrays. For this restriction, there exists no substantial reason — except that for a normal C++ function, we have to fix a type for each argument.

Type parameters

Now, if an argument type could itself be a parameter, then we could implement sum once for all value types of the array argument. Such a type parameter can be specified using the C++ template syntax:

template<typename T>
T sum1(T* a, int n)
{
  T res = 0;
  for(int i = 0; i < n; ++i)
    res += a[i];
  return res;
}

The parameter T takes over the role of the previously constant type double, and thus our call above now works:

double a[10];
float  b[10];
int    c[10];
// ...
double sa = sum1(a, 10); // OK
float  sb = sum1(b, 10); // OK
int    sc = sum1(c, 10); // OK

So, we can process array with arbitrary value type (int, std::complex<double>, myvector ... if implemented reasonably), or even std::string (well, almost ... we'll come back to that). So, are we now entirely happy?

• <<<Previous page
• Next page >>>