Question

I am trying to pass a vector of vectors by reference. I have typdefed the datatype and it seems to me that I am getting a copy, not a reference. I couldn’t find figure out any valid syntax to do what I want here. Suggestions?

#include <iostream>
#include <fstream>
#include <vector>
#include <string>
#include <cmath>

#define __DEBUG__

using namespace std;

//Define custom types and constants
typedef std::vector< std::vector<float> > points;
//Steup NAN
float NaN = 0.0/0.0;  //Should be compiler independent

//Function prototypes
void vectorFunction(float t0, float tf,  points data );

//Global constants
string outFilename = "plotData.dat";
int sampleIntervals = 10000; //Number of times to sample function.

int main()
{
    ofstream plotFile;
    plotFile.open(outFilename.c_str());

    points data;

    vectorFunction( 0, 1000, data );

#ifdef __DEBUG__
    //Debug printouts
    cout << data.size() << endl;
#endif

    plotFile.close();
    return 0;
}

void vectorFunction(float t0, float tf, points data )
{
    std::vector< float > point(4);
    float timeStep = (tf - t0)/float(sampleIntervals);
    int counter = floor(tf*timeStep);

    //Resize the points array once.

    for( int i = 0; i < counter; i++)
    {
        point[0] = timeStep*counter;
        point[1] = pow(point[0],2);
        point[2] = sin(point[0]);
        point[3] = -pow(point[0],2);
        data.push_back(point);
    }

#ifdef __DEBUG__
    //Debug printouts
    std::cout << "counter: " << counter
              << ", timeStep: " << timeStep
              << ", t0: " << t0
              << ", tf: " << tf << endl;
    std::cout << data.size() << std::endl; 
#endif

}

void tangentVectorFunction(float t0, float tf, points data)
{

}

Answer 1

Assuming your typedef remains:

typedef std::vector< std::vector<float> > points;

Your prototype for passing by reference would look like this:

void vectorFunction(float t0, float tf, points& data);
void tangentVectorFunction(float t0, float tf, points& data);

Your points type is just a value type, and is equivalent to std::vector< std::vector<float> >. Assignment to such a variable makes a copy. Declaring it as a reference type points& (or std::vector< std::vector<float> >&) uses a reference to the original.

It certainly doesn’t make a difference in the scope of your problem, but you might consider simply using a one dimensional vector. You save a little bit on memory allocations, deallocations, and lookups this way. You’d use:

point_grid[width * MAX_HEIGHT + height] // instead of point_grid[width][height]