Question

My question is related to numeric type conversion in C++. A very common way to do that is to use static_cast, for example:

float a;
int b;
a = 3.14;
b = static_cast<int>(a);

Then, how about numeric vector type conversion? Could we continue to use static_cast? I have done the following experiment:

typedef vector<int> IntVector;
typedef vector<float> FloatVector;
IntVector myvector;
myvector.push_back(3);
myvector.push_back(4);
myvector.push_back(5);

// Solution 1 (successful)
FloatVector solution1 ( myvector.begin(), myvector.end() );
for(int i=0; i<solution1.size(); i++)
   cout<<solution1[i]<<endl;
// Solution 2 (failed)
FloatVector solution2;
solution2 = static_cast<FloatVector> (myvector);

It seems that for numeric vector types it is impossible to use static_cast to convert. I was wondering whether there are good solutions to this problem. Thanks!

Answer 1

The language directly supports conversion from one numeric type to another. You do not even need the static_cast, you could just assign. This conversion involves a logical copying of the value, as opposed to a reinterpretation of the value representation.

The language does not directly support conversion between arrays of different types, or for that matter of std::vector of different types.

But as you found, there is some support for copying elements, and then when each element is numeric, the built-in support for numeric type conversion kicks in for each element.