The best way to free the memory in static function - c++

Question

This is a Matrix implementation which is working now.
And I need to deallocate the memory of the *data .
But the readMatrix function is static, so I cannot delete [] the data in readMatrix.
What is the best way to do it after the readMatrix returned the *m matrix.
Matrix.cpp
#include "Matrix.h"
#include <sstream>
#include <fstream>
#include <iostream>
#include <string>
using namespace std;
Matrix::Matrix(int i, int j) {
rows=i;
cols=j;
data = new float [i * j];
}
Matrix::Matrix(const Matrix& m) {
rows=m.rows;
cols=m.cols;
data=m.data;
}
int Matrix::numRows() {
return rows;
}
int Matrix::numCols() {
return cols;
}
float *Matrix::access(const int i, const int j) const {
return &data[(((i * cols) + j) - 1)] ;
}
std::ostream& operator<<(std::ostream &os, Matrix &m) {
int i, j;
os << m.numRows() << " " << m.numCols() <<endl;
for (i = 0; i < m.numRows(); i++) {
for (j = 0; j < m.numCols(); j++) {
os << *(m.access(i, j)) << " ";
}
os << endl;
}
return os;
}
int **Create2D(int row, int col)
{
int **p = new int* [row];
for (int j = 0; j < row; j ++)
p[j] = new int[col];
return p;
}
// Deletes an array pointed by 'p' that has 'row' number rows
void Delete2D(int **p, int row)
{
for (int j = 0; j < row; j ++)
delete [] p[j];
delete [] p;
}
Matrix Matrix::readMatrix(std::string filename)
{
int r ,c;
ifstream matrixFile(filename.c_str());
matrixFile >> r >> c;
int **p = Create2D(r, c);
Matrix* m = new Matrix(r, c);
for (int i=0; i<r; i++) {
for (int j=0; j<c; j++) {
matrixFile >> p[i][j];
*(m->access(i,j))= (float)p[i][j];
}
}
matrixFile.close();
Delete2D(p, r);
return *m;
}
Matrix.h
#ifndef MATRIX_H
#define MATRIX_H
#include <stdlib.h>
#include <iostream>
#include <fstream>
class Matrix {
public:
Matrix(int i, int j) ;
Matrix (const Matrix& m) ;
int numRows ( ) ;
int numCols ( ) ;
float *access(const int i, const int j) const ;
friend std::ostream& operator<<(std::ostream &os, Matrix &m) ;
static Matrix readMatrix ( std::string filename ) ;
private:
Matrix() { }
// ~Matrix() { }
int rows ;
int cols ;
float *data ;
} ;
#endif // MATRIX_H
Thank you!

Answer 1

This here
Matrix::Matrix(const Matrix& m) {
rows=m.rows;
cols=m.cols;
data=m.data;
}
will not work well with your bare float pointer (note that it should be a float**), you end up having more instances pointing to the same data and when they go out of scope they will be deleting the same data more than once causing UB.
What you need to do in a copy constructor is to clone the data. Since you have the rows / cols it should be fairly easy.
now to your question.
This here
...
matrixFile >> p[i][j];
*(m->access(i,j))= (float)p[i][j];
...
(note that you declared access with const at the end (float *access(const int i, const int j) const), this means that the above assignment is illegal, const after the method means the instance will not be affected but you are using it to assign it a value)
seems unnecessary since you are reading one float at a time, it would be enough just to have a single float variable which you use to read the value from the file then assign your matrix that value
...
matrixFile >> floatRead;
*(m->access(i,j))= floatRead;
...
so just ditch the whole p matrix.
Alt. read in the values in the p matrix then let the Matrix instance m take ownership of it.
e.g.
...
m->attachMatrixValues(p);
...