With C arrays, it is fairly easy to write code that takes arrays of any size:
void func( T* itBegin, T* itEnd );
void main() {
T arr1[1];
func( std::begin(arr1), std::end(arr1) );
T arr2[2];
func( std::begin(arr2), std::end(arr2) );
}
How can I do that with std::arrays?
void func( ??? itBegin, ??? itEnd );
void main() {
std::array<T,1> arr1;
func( std::begin(arr1), std::end(arr1) );
std::array<T,2> arr2;
func( std::begin(arr2), std::end(arr2) );
}
The problem is that, in MSVC 2010, std::array<T,N>::iterator is different for different N. Is this a bug in MSVC 2010? If not, what is the rationale of this design? Yes, I could get pointers from the std::array and pass them instead of iterators, but isn't that unnecessarily ugly?
BTW, boost::array<T,N>::iterator are the same for all N.
template <class I>
void func(I begin, I end)
{
for (auto x = begin; x != end; ++x)
something_with(*x);
}
Define them genericly as a type parameter, and then just use them as if they were pointers. Anything that behaves pointer-like will compile, things that don't, won't.
Pointer-like things include normal pointers, as well as standard library iterators, and anything else that defines operator=, operator* and operator++.
Doing it this way and as you will only ever use matching pairs of begin/end iterator ranges from the same array<N>, then it doesn't matter if array<N>::iterator is a different type to array<M>::iterator.
As far as I can tell, the standard doesn't require that different sized std::array have the same type of iterator; having a different type for std::array<int, 1> and std::array<int, 2> seems legal (although one might have some opinions with regards to the quality of the implementation).
If this is a problem, you can either use a C style array, or use pointers:
func( &arr1[0], &arr1[0] + arr1.size() );
Neither solution is ideal, but they're the best I can offer.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With