I have a template function which does something with a given value as long as the template type is a number:
template <typename scalar_t>
void function(scalar_t value)
{
if constexpr (std::is_floating_point_v<scalar_t>)
{
std::cout << value << 'F';
}
if constexpr (std::is_integral_v<scalar_t>)
{
std::cout << value;
}
}
And an overload for this template expecting a container containing numbers:
template <typename scalar_t, template <typename> class container_t>
void function(const container_t<scalar_t> &container)
{
for (const auto &value : container)
{
function(value);
std::cout << ' ';
}
}
With the templates above, I call the following code:
int main()
{
function('0');
function(short{1});
function(2);
function(3l);
function(4ll);
function(5.f);
function(6.);
std::vector<int> v{7, 8, 9, 10};
std::list<double> l{11., 12., 13., 14.};
function(v);
function(l);
return 0;
}
Which gives the following output:
012345F6F
While I was expecting the following output:
012345F6F7 8 9 10 11F 12F 13F 14F
Which means that the first template version (function(scalar_t value)) has been chosen instead of the second version (function(const container_t<scalar_t> &container)).
How can I force the second version of the template-function to be chosen for template template parameters?