I have a C++ design problem regarding the base class and the derived class. Here is the code looks like:
class BaseClass:
{
public:
BaseClass();
virtual ~BaseClass();
//...Some basic functions for the derived Classes
void write_data(void* data);
};
class DerivedClass1 : BaseClass
{
public:
DerivedClass1(){}
~DerivedClass1() override{}
void method(Type1 data);
}
class DerivedClass2 : BaseClass
{
public:
DerivedClass2(){}
~DerivedClass2() override{}
void method(Type2 data);
}
// There are hundreds of auto-created derived classes
.....
The class that instantiates the derived classes uses a vector to store the instantiated derived classes:
std::vector<std::unique_ptr<BaseClass>> m_factory_vector;
The function that stores the classes is:
template<typname Type>
void fun(Type& data)
{
std::unique_ptr<BaseClass> base= std::make_unique<Type>(*file_);
base.get()->method(data);
m_factory_vector.push_back(base);
}
The reason that using the template is due to the number of functions call the fun is unknown.
Now I would like to iterate through the vector and call the method in the derived classes again. Because the methods write a different type of data into the HDF5 file by collecting the data in a buffer(size 100, but the total size is unknown), but there is some data remaining in the collected batch that I would like to flush when the program is ended. Here is the problem I am having, I wish I can do it like:
for(hsize_t i=0;i<m_factory_vector.size();i++)
{
m_factory_vector[i]->method({});
}
But I know the problem is the method is not in the BaseClass but in the derived classes. But the method in derived classes has different types of input parameters that I may not able to declare virtual void method() in BaseClass. However, I still want to have something like this so that I can call flush those data into the database without losing the data.