You will have to define the type yourself, it will probably be something like the following:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
class only_int { int val_;
public: only_int (int v = 0) : val_(v) {}
template <typename T, ENABLE_IF(is_signed_integral<T>)> only_int (T v) : val_(v) {} // can static_assert the size
template <typename T, ENABLE_IF(!is_signed_integral<T>)> only_int (T) = delete;
int get() const { return val_; } }; |
is_signed_integral is a type trait I have defined for describing signed integral types:
|
1 2 3 4 5 6 |
template <typename T> using is_signed_integral = typename std::conditional< std::is_signed<T>::value && std::is_integral<T>::value, std::true_type, std::false_type >::type; |
ENABLE_IF is a macro I defined to make the code that uses enable_if tricks more concise. This syntax works in C++11, and the macro is defined as:
|
1 2 3 |
# define ENABLE_IF(...) \ typename std::enable_if<__VA_ARGS__::value, bool>::type \ = true |
The idea behind the three constructors of only_int is the following. The first constructor simply works when an int is used for copy construction. The second and third constructor declarations provide a ‘conditional’ constructor definition. If condition is_signed_integralis satisfied, the object is properly initialized. Otherwise, e.g., when copy-initializing from adouble constructor is declared as deleted: a compile-time error is triggered.
