I'm writing a vector template with a few specializations so that there are named accessors for the first few items of the vector (such x, y, z, w) depending on the size. KennyTM's answer was a great help with this. I also added a form of Curiously recurring template pattern to be able to use the operators naturally. Here is the base class and a specialization with added functions (omitted most member implementations for brevity):
template<int Dim, typename V, typename ItemType = float>
class VectorImpl
{
public:
typedef ItemType value_type;
VectorImpl() { elements.fill(0); }
VectorImpl(std::initializer_list<ItemType> init_list);
V operator+(const V& other)
{
V result;
for (int i = 0; i < Dim; ++i)
result.elements[i] = elements[i] + other.elements[i];
return result;
}
QString toString();
// ... other members ...
protected:
VectorImpl(const VectorImpl& other) = default;
protected:
std::array<ItemType, Dim> elements;
};
template<int Dim, typename ItemType = float>
class Vector : public VectorImpl<Dim, Vector<Dim, ItemType>, ItemType>
{
typedef VectorImpl<Dim, Vector<Dim, ItemType>, ItemType> ParentType;
public:
Vector() : ParentType() {}
Vector(const ParentType& other) : ParentType(other) {}
Vector(std::initializer_list<ItemType> init_list) : ParentType(init_list) {}
};
template<typename ItemType>
class Vector<3, ItemType> : public VectorImpl<3, Vector<3, ItemType>, ItemType>
{
typedef VectorImpl<3, Vector<3, ItemType>, ItemType> ParentType;
public:
Vector() : ParentType() {}
Vector(const ParentType& other) : ParentType(other) {}
Vector(std::initializer_list<ItemType> init_list) : ParentType(init_list) {}
ItemType x() const { return this->elements[0]; }
ItemType y() const { return this->elements[1]; }
ItemType z() const { return this->elements[2]; }
};
And I wanted to add qDebug()<<
support, so I did this:
template<int Dim, typename ItemType>
QDebug operator<<(QDebug dbg, Vector<Dim, ItemType>& v)
{
dbg.nospace() << v.toString();
return dbg.space();
}
Now, the following code compiles and works:
Vector<3> v1 = { 3,4,5 };
qDebug() << v1;
This one does, too:
Vector<3> v1 = { 3,4,5 };
Vector<3> v2 = { 1,-1,1 };
qDebug() << v1;
auto v3 = v1 + v2;
qDebug() << v3;
But this one does not:
Vector<3> v1 = { 3,4,5 };
Vector<3> v2 = { 1,-1,1 };
qDebug() << (v1 + v2);
The compiler says:
error: no match for 'operator<<' in 'qDebug()() << v1.Vector<3>::.VectorImpl::operator+ [with int Dim = 3, V = Vector<3>, ItemType = float]((*(const Vector<3>*)(& v2)))'
What is going on? Why is the type of v1 + v2
different when assigning to a variable? What should I do to make this compile?