C++ noob: proper way to avoid redundant object copies when initializing? [duplicate] - c++

Question

This question already has answers here:
What is The Rule of Three?
(8 answers)
Closed 8 years ago.
I made a minimal example:
#include <iostream>
#include <conio.h>
using namespace std;
// skipped getters and setters and bounds checking for brevity
struct Vertex {
int x,y;
Vertex() {
}
Vertex(int x, int y) {
this->x = x;
this->y = y;
}
};
struct Polygon {
Vertex *vertexlist;
Polygon() {
}
Polygon(Vertex *v) {
vertexlist = new Vertex[4]; //hard coded 4 vertices for example brevity
for(int i=0;i<4;i++) {
vertexlist[i] = v[i];
}
}
Vertex& getVertex(int index) const {
return this->vertexlist[index];
}
};
struct PolyList {
Polygon *polylist;
int lastpoly;
PolyList() {
polylist = new Polygon[10]; //hard coded 10 for example brevity
lastpoly = 0;
}
void add(const Polygon& p) {
polylist[lastpoly++] = p;
}
};
ostream& operator<<(ostream& o, Vertex& v) {
return o << "(" << v.x << ", " << v.y << ")";
}
ostream& operator<<(ostream& o, const Polygon& p) {
for(int i=0;i<4;i++) {
o << p.getVertex(i) << ", ";
}
return o;
}
ostream& operator<<(ostream& o, PolyList& pl) {
for(int i=0;i<pl.lastpoly;i++) {
o << pl.polylist[i] << endl;
}
return o;
}
int someFunc() {
Vertex *vl = new Vertex[4];
PolyList pl;
vl[0] = Vertex(1,2);
vl[1] = Vertex(3,4);
vl[2] = Vertex(5,6);
vl[3] = Vertex(7,8);
pl.add(Polygon(vl)); // this Polygon goes out of scope after this line
cout << pl << endl;
}
int main() {
someFunc();
}
(So tl;dr, Polygon is a list of 4x Vertex, and PolyList is a list of Polygon:s. Polygon:s are add()ed to PolyList by instantiating a temporary Polygon)
Now, this leaks memory, because the Vertices in Polygon are never freed. However, if I add a destructor:
Polygon::~Polygon () {delete [] vertices}
then
cout << pl << endl;
will not work because the Polygon has gone out of scope and the destructor frees the vertices.
I could have the PolyList destructor call a Polygon->free() function. Alternatively, I could have the Polygon::Polygon(Vertex *v) deep copy all the vertices in v. Then PolyList::PolyList(Polygon &p) could deep copy p.
I could also make a PolyList::createPolygon(int x1, int y1, int x2, int y2...) but that flies in the face of OO.
What is the proper way to handle this kind of situation in C++? Never mind my actual example where a memory leak would not be a problem, I'm talking in principle. If I make an hierarchical object tree, I want to copy the pointers, not deep copy the objects.
EDIT: I'm trying to learn C++ on a deep level, so this is not about using vector<> or another "canned solution"; that is not what I'm after here, though I'm sure that is a good solution if the above example was an actual problem I was having. The example above is just the briefest example I could think of to explain my question.

Answer 1

You could use smart pointers and STL containers (mainly std::vector as suggested by PaulMcKenzie).
They will help a lot.
Your example using std::vector
#include <iostream>
#include <conio.h>
#include <vector>
using namespace std;
// skipped getters and setters and bounds checking for brevity
struct Vertex {
int x, y;
Vertex() {
}
Vertex(int x, int y) {
this->x = x;
this->y = y;
}
};
typedef vector<Vertex> vertex_list_t;
struct Polygon {
vertex_list_t vertexlist;
Polygon() {
}
Polygon(vertex_list_t v) {
//hard coded 4 vertices for example brevity
for (int i = 0; i<4; i++) {
vertexlist.push_back(Vertex(i, i));
}
}
Vertex getVertex(int index) const {
return vertexlist[index];
}
};
typedef vector<Polygon> polygon_list_t;
ostream& operator<<(ostream& o, Vertex& v) {
return o << "(" << v.x << ", " << v.y << ")";
}
ostream& operator<<(ostream& o, const Polygon& p) {
for (auto v: p.vertexlist) {
o << v << ", ";
}
return o;
}
ostream& operator<<(ostream& o, polygon_list_t& pl) {
for (auto &p : pl) {
o << p << endl;
}
return o;
}
int someFunc() {
vertex_list_t vl = {
Vertex(1, 2),
Vertex(3, 4),
Vertex(5, 6),
Vertex(7, 8)
};
polygon_list_t pl;
pl.push_back(Polygon(vl)); // this Polygon goes out of scope after this line
cout << pl << endl;
return 0;
}
int main() {
someFunc();
}
What's the real deal?
In the line
pl.add(Polygon(vl)); // this Polygon goes out of scope after this line
you pass the polygon as a temporary and:
$12.2/3- "Temporary objects are destroyed as the last step in evaluating the full-expression (1.9) that (lexically) contains the point where they were created. This is true even if that evaluation ends in throwing an exception."
change that line by:
Polygon p1(vl);
pl.add(p1); // this Polygon NOT goes out of scope after this line

Answer 2

You can use shared_ptrs as the solution. i.e.
#include "stdafx.h"
#include <iostream>
#include <conio.h>
#include <memory>
#include <list>
#include <vector>
using namespace std;
struct Vertex
{
int x,y;
Vertex() : x(0), y(0)
{
}
Vertex(int _x, int _y)
{
x = _x;
y = _y;
}
};
struct Polygon
{
vector<Vertex> vertexes;
Polygon()
{
}
Polygon(Vertex *v)
{
const int ELEMS_COUNT = 4;
vertexes.reserve(ELEMS_COUNT);
vertexes.insert(vertexes.end(), v, v + ELEMS_COUNT);
}
Vertex getVertex(int index) const
{
return vertexes[index];
}
};
typedef shared_ptr<Polygon> PolygonPtr;
struct PolyList
{
std::list<PolygonPtr> polylist;
void add(PolygonPtr p)
{
polylist.push_back(p);
}
};
ostream& operator<<(ostream& o, const Vertex& v) {
return o << "(" << v.x << ", " << v.y << ")";
}
ostream& operator<<(ostream& o, const Polygon& p) {
for (auto& p : p.vertexes)
{
o << p << ", ";
}
return o;
}
ostream& operator<<(ostream& o, PolyList& pl) {
for(auto& p : pl.polylist)
{
o << *p << endl;
}
return o;
}
int someFunc() {
Vertex vl[] = {Vertex(1, 2), Vertex(3, 4), Vertex(5, 6), Vertex(7, 8)};
PolyList pl;
pl.add(PolygonPtr(new Polygon(vl)));
cout << pl << endl;
return 0;
}
int main()
{
someFunc();
return 0;
}