Is there anything wrong with using C++ vectors in structures?

Question

Ok, I'm coming from vb.net to c++. Im trying to use vectors in a structure, but the compiler yells at at me for it. What is wrong with the current statement?

#include <vector>

struct FactorSet
{
 vector<long long> UpperFactor(0);
 vector<long long> LowerFactor(0);
};

Output error (Visual Studio 2008):

Error 1 error C2059: syntax error : 'constant'

I venture to guess it is my lack of understanding of what a vector really is. In my mind it is an object, although I think its whats called a template. Other objects like strings seem to have no problem. I also assume this is extended to class definitions as well since structures and classes are so similar.

Answer 1

You want:

#include <vector>

struct FactorSet
{
 std::vector<long long> UpperFactor;
 std::vector<long long> LowerFactor;
};

though you might also have problems with long long, as this is not currently part of C++.

If you actually want to give a size to the vectors, you need to do that via the structs constructor:

struct FactorSet
{
 std::vector<long long> UpperFactor;
 std::vector<long long> LowerFactor;

 FactorSet() : UpperFactor(42), LowerFactor(42) {}
};

So now when you say:

FactorSet f;

the vectors in f will both have size 42.

As to what a vector actually is, it's a class very like a string, except in the case of the vector you have to say what type of thing it contains. So

vector <char> s;

is very similar (but not identical) to:

string s;

Answer 2

Yes, you can use the vector in a struct or class. The problem is that in C++, you cannot initialize members inline; you need to do initialization in the constructor. For example:'

class FactorSet {
   public:
       FactorSet() : UpperFactor(0), LowerFactor(0) {}
       // ...
   private:
       std::vector<int64_t> UpperFactor;
       std::vector<int64_t> LowerFactor; 
};

Just some comments... vector will, by default, be constructed with zero elements, so it is not necesary to explicitly construct it that way (in fact, doing so might be slightly slower than just using the default constructor). Also, the type long long is presently non-standard. Consider using int64_t defined in stdint.h (and, more portably, in the header boost/cstdint.hpp).