Memory overwrite in my own Vector class - c++

For my Algorithm course project we can't use STL stuff like std::vector and so I'm trying to implement my own version of it (with templates).
It seems it works but when I declare a Vector< Vector< int > >
the .push() method starts to overwrite memory.
More specifically, with this code:
Vector<Vector<int>> v(3);
cout << v[0].push(0) << "\n";
cout << v[0].push(55) << "\n";
cout << v[0].push(4) << "\n";
cout << v[1].push(12) << "\n";
cout << v[1].push(3) << "\n";
cout << v[2].push(1) << "\n";
The output is this (.push() returns the address of where the element is inserted):
0x561328b0bc20
0x561328b0bc24
0x561328b0bc28
0x561328b0bc20
0x561328b0bc24
0x561328b0bc20
Any suggestion of why this happens?
Here is the code for my Vector class:
#include <iostream>
#include <bits/stdc++.h>
using namespace std;
template<class T>
class Vector {
private:
size_t _size;
size_t _capacity;
char* _buffer; //char* for performance
void _realloc(size_t);
public:
Vector(size_t s=0, T def=T());
T* push(T);
T& operator[](int);
size_t size() { return _size; }
};
template<class T>
void Vector<T>:: _realloc(size_t ncap)
{
_capacity = ncap;
char* nbuf = _buffer;
_buffer = new char[_capacity];
for(size_t i=0; i<_size * sizeof(T); ++i)
_buffer[i] = nbuf[i];
delete[] nbuf;
}
/*
* s -> size
* def -> default value
*/
template<class T>
Vector<T>:: Vector(size_t s, T def) : _size(s)
{
_capacity = 32;
while(_capacity < _size)
_capacity *= 2;
_buffer = new char[_capacity * sizeof(T)];
for(size_t i=0; i<_size; ++i)
((T*)_buffer)[i] = def;
}
/*
* check capacity, reallocs if necessary and push the element
* then return the addres (used only for debug)
*/
template<class T>
T* Vector<T>:: push(T ele)
{
if(_capacity == _size)
_realloc(2 * _capacity);
((T*)_buffer)[_size++] = ele;
return &((T*)_buffer)[_size-1];
}
template<class T>
T& Vector<T>:: operator[](int i)
{
if(i<0 or i>=(int)_size) {
cerr << "Out of bounds!\n";
abort();
}else
return ((T*)_buffer)[i];
}
template<class T>
ostream& operator<<(ostream& out, Vector<T>& v)
{
out << "{";
if(v.size() > 0) {
out << v[0];
for(size_t i=1; i<v.size(); ++i)
out << ", " << v[i];
}
out << "}";
return out;
}
Thanks!
PS: I know it's not a good use of C++ :P

Your operator= implicitly defined does the wrong thing. You use it in your constructor.
So, follow the rule of 0/3/5: Implement copy/move construtors/assignment and destructors, as this is an owning memory-management type. (Non-resource management types should follow the rule of 0; copyable resource management types the rule of 5.)
See rule of three. Implement all of destructor/copy constructor/move constructor/copy assign/move assign, or none of them.
Don't copy the data byte-wise when you realloc. std::move the Ts from the source to the dest.
In the copy construct/assign you'll want to copy the source Ts, not the underlying bytes.

Related

C++: Disassemble a flat vector into multiple vectors of equal size without copying

Is it possible in C++ to split a flat vector (or C style array) into multiple vectors of equal size without copying any of its containing data? That is, disassembling the original vector by moving its content to a new vector, which invalidates the original vector. The following code example should illustrate this:
#include <cassert>
#include <vector>
void f(int* v) {
for (int i = 0; i < 100; i++) {
v[i] = i;
}
}
/**
* Split v into n vectors of equal size without copy its data (assert v.size() % n == 0)
*/
std::vector<std::vector<int>> g(std::vector<int> v, int n) {
std::vector<std::vector<int>> vs(n);
int vec_size = v.size() / n;
for (int i = 0; i < n; i++) {
vs[i].assign(v.begin() + i * vec_size, v.begin() + (i + 1) * vec_size); // copies?
// how to let vs[i] point to v.begin() + i * vec_size?
}
return vs;
}
int main() {
std::vector<int> v(100);
f(v.data());
std::vector<std::vector<int>> vs = g(std::move(v), 10);
for (int i = 0; i < 10; i++) {
for (int j = 0; j < 10; j++) {
assert(vs[i][j] == i * 10 + j);
}
}
return 0;
}
Yes, in my opinion this is possible. Moving the elements, but not copying the elements.
C++ offers std::make_move_iterator. Please read here about that.
To check that, I created a small class to output, to see, if we copy or move something.
So, if your data can "move", then it will work, otherwise of course a copy will be made. With the following we see the result.
struct Test {
int data{};
Test(int d) : data(d) { std::cout << "Construct and init\n"; }
Test() { std::cout << "Default construct\n"; };
~Test() { std::cout << "Destruct\n"; };
Test(const Test& other) { std::cout << "Construct\n"; data = other.data; }
Test(const Test&& other) noexcept { std::cout << "Move Construct\n"; data = other.data; }
Test& operator =(const Test& other) noexcept { std::cout << "Assign\n"; data = other.data; return *this; }
Test& operator =(const Test&& other) noexcept { std::cout << "Move Assign\n"; data = other.data; return *this; }
};
We will additionally add a small function, which calculates the offsets of the chunks that will be moved.
And then, we can come up with a small function to implement that.
#include <iostream>
#include <vector>
#include <numeric>
#include <iterator>
#include <iomanip>
// Calculate start and end index for all chunks
std::vector<std::pair<size_t, size_t>> calculatePairs(const size_t low, const size_t high, const size_t numberOfGroups) {
// Here we will store the resulting pairs with start and end values
std::vector<std::pair<size_t, size_t>> pairs{};
// Calculate chung size and remainder
const size_t delta = high - low;
const size_t chunk = delta / numberOfGroups;
size_t remainder = delta % numberOfGroups;
// Calculate the chunks start and end addresses for all chunks
size_t startIndex{}, endIndex{};
for (size_t i = 0; i < numberOfGroups; ++i) {
// Calculate end address and distribute remainder equally
endIndex = startIndex + chunk + (remainder ? 1 : 0);
// Store a new pair of start and end indices
pairs.emplace_back(startIndex, endIndex);
// Next start index
startIndex = endIndex;
// We now consumed 1 remainder
if (remainder) --remainder;
}
//--pairs.back().second;
return pairs;
}
struct Test {
int data{};
Test(int d) : data(d) { std::cout << "Construct and init\n"; }
Test() { std::cout << "Default construct\n"; };
~Test() { std::cout << "Destruct\n"; };
Test(const Test& other) { std::cout << "Construct\n"; data = other.data; }
Test(const Test&& other) noexcept { std::cout << "Move Construct\n"; data = other.data; }
Test& operator =(const Test& other) noexcept { std::cout << "Assign\n"; data = other.data; return *this; }
Test& operator =(const Test&& other) noexcept { std::cout << "Move Assign\n"; data = other.data; return *this; }
};
std::vector<std::vector<Test>> split(std::vector<Test>& v, unsigned int n) {
std::vector<std::vector<Test>> result{};
if (v.size() > n) {
result.resize(n);
std::vector<std::pair<size_t, size_t>> offset = calculatePairs(0u, v.size(), n);
for (size_t i{}; i < n; ++i) {
result[i].insert(result[i].end(), std::make_move_iterator(v.begin() + offset[i].first),
std::make_move_iterator(v.begin() + offset[i].second));
}
}
return result;
}
constexpr size_t NumberOfElements = 30u;
constexpr unsigned int NumberOfGroups = 3;
static_assert (NumberOfElements >= NumberOfGroups, "Number of elements must be greater/equal then number of elements\n");
int main() {
std::cout << "\n\n\nCreate vector with " << NumberOfElements << " elements\n\n";
std::vector<Test> v1(NumberOfElements);
std::cout << "\n\n\nFill vector with std::iota\n\n";
std::iota(v1.begin(), v1.end(), 1);
std::cout << "\n\n\nSplit in " << NumberOfGroups<< "\n\n";
std::vector<std::vector<Test>> s = split(v1, NumberOfGroups);
std::cout << "\n\n\nOutput\n\n";
for (const std::vector<Test>& vt : s) {
for (const Test& d : vt) std::cout << std::setw(3) << d.data << ' ';
std::cout << "\n\n";
}
}
But my strong guess is that you want to splice the data. The underlying elements fo the std::vector which you can get with the data() function.
You can access the data easily with pointer arithmetic on data().
But if you want to have the data in a new container, then this is difficult with a std::vector. It can for example be done with a std::list that has a splice function and does, what you want.
Or, you need to implement your own dynamic array and implement a splice function . . .
Checksum:
;23M#eTo1?:B#r7C8#wtJ'Z'..uIvLT.j;bld$Bvgjd.qm=8;B/`dHM%D#wyv:\5YI:WVGwJL00%IsKQ9O+&#g,/gzkPg^cg::LX?6dL3;Fs3GOOAmQmCIW?&skWxZXsElyn6S3#fi:0DSKJ/A^r#*'k#a#e8!XDpjAUtu?K5iu+e=P"M7a2BWdFdA.5NP:Y"l,,h+Y/PxhVfP/m0ceS=Nxol2vOZwM2+!H\^a=douX%fhqcr4'0eXiEZeKvTf0^%CTNY^WB6fc#IpK^GQgxTXQo0ikr0+/OxXlc1B$5jc1r,GQj+fwEdoCPrz6,j:SO6L3QU#7lT:f#Y^V!Au\P'a5amR$NCU?\WspBOuy#RH3tJimka#rdyNN56.$;DtRCHN*YeWlrG=',XNSrzEK:Cw;#A%.#/:c,a2W24IIIdecc7O"EnKQn;nXmUemX4kclDsYci+izmr#vlGAQ.w2!cuf;6n2UvJM,CeSyRj1,:2\9#i8GLwtux!uEHUp7X*5SC%nld956CHsy&/n73/90cRP'Me"1PW+##FH8mH4Rf^o=ZP/Rm\X&1syUdUh+.N/jtoO:,OBBAmq,jW69Fu%jJukBa$g4hIrIPcxx17i;XU,FCbQGd8v*AyKGSML\JN#jte*F";Zh7fqhvCXobE&SapX90r"Z$.CN,1R^aj.=5L6^tUB2UPJH^eb'*B!v5=D.9PFI#Pt*KjK+yS*tV6f.5kgPOzBE$uK0MA/\l9U"63LUR6k3#'cub?u&xILMXP%#:lx2TbKhFOjBpMN!+%F16jrgv&AoFhuf%P!==8?x,NsSd%hVo"BJhVv3rjrhvM"WLE3%y#N7g37Re^XiS9lpyKA9E7ow6U=I"tlv",&#+fZoIR4KM58!NTm978wCI?9wo.ocS!9i5k#ler47J.G0yXjZVSdr=G"uRodC06k\V%8;oUwV&z!W5:+ZvE:nyO#+lO+Hn0&tnH&^tNC?'PmERxs/B+KW4O6&oWDED9?MqxmYgVKoT.a%iw

What's the suited container to push values on top, remove at whatever index and avoid memory reallocation?

I need to build a sort of stack where I can push values on top:
5 // (size 1)
5 3 // (size 2)
5 3 8 // (size 3)
than remove them by value, such as removing 3:
5 8 // (size 2)
than be able to always get the last value (i.e. 8 in the example), when I need it).
I can push max 32 values, so I know the whole size (avoiding heap?).
I think to std::vector with:
initial reserve(32)
.push_back() for insert
vector.erase(std::remove(vector.begin(), vector.end(), value), vector.end()) for remove by value
vector[vector.size() - 1] to retrieve the last element
But maybe there are some stl container better for this kind of process? Not sure if vector are always in the stack and will do further memory reallocation under the hood...
You can write an allocator that contains your 32 values, and refuses to allocate any amount other than 32
template <typename T, std::size_t N = 32>
struct static_allocator
{
T* allocate(std::size_t n) { if (n != N) throw std::bad_alloc(); return arr; }
void deallocate(T *, std::size_t) {}
using pointer = T*;
using const_pointer = const T*;
using void_pointer = void*;
using const_void_pointer = const void*;
using value_type = T;
using size_type = std::size_t;
using difference_type = std::ptrdiff_t;
template <typename U>
struct rebind
{
using other = static_allocator<U, N>;
};
static_allocator select_on_container_copy_construction() { return {}; }
using propagate_on_container_copy_assignment = std::true_type;
using propagate_on_container_move_assignment = std::true_type;
using propagate_on_container_swap = std::true_type;
private:
T arr[N];
};
Then a std::vector<T, static_allocator<T>> will have it's elements as subobjects.
I don't think it's possible to avoid dynamic allocation and have sublinear random-access remove.
if size is limited to 32 elements
why not use a circular buffer of 32 elements, and roll the elements when they are 32 ?
There may be some bugs (don't use last() or remove () on an empty container, don't remove an element not inserted...) ,but it works for the functions you wanted. Here is the idea (heap is avoided)
#include <iostream>
template <typename T>
class Container {
public :
static const int smax_ = 32;
void erase () {
T* pt ((T*) val_);
for (int i (0); i != smax_; ++i, ++pt) *pt = 0;
size_ = 0;
}
Container () : size_ (0) { erase ();}
~Container () {}
void copy (const Container& c) {
size_ = c.size_;
T* pt ((T*) val_);
const T* qt ((const T*) c.val_);
for (int i (0); i != size_; ++i, ++pt, ++qt) *pt++ = *qt++;
}
Container (const Container& c) {
copy (c);
}
void push_back (const T& t) {
if (size_ == smax_) {
T* pt ((T*) val_);
const T* qt ((const T*) val_);
++qt;
for (int i (0); i != size_ -1; ++i, ++pt, ++qt) {
*pt = *qt;
}
*pt = t;
}
else {
val_ [size_] = t;
++size_;
}
}
int size () const {
return size_;
}
void remove (const T& t) {
if (!size_) return;
int i (0);
T* pt ((T*)val_);
while ((i < smax_) && (*pt != t)) {
++pt; ++i;
}
if (i != smax_) {
T* qt (pt);
++qt;
for (; i != size_ -1; ++i, ++pt, ++qt) {
*pt = *qt;
}
}
--size_;
}
void write (std::ostream& os) const {
const T* pt ((const T*) val_);
for (int i (0); i != size_; ++i, ++pt) os << *pt << " ";
}
bool operator == (const Container& c) const {
if (size_ != c.size_) return false;
const T* pt ((const T*) val_), *qt ((const T*) c.val_);
for (int i (0); i != size_; ++i, ++pt, ++qt) if (*pt != *qt) return false;
return true;
}
bool operator != (const Container& c) const {
return !operator == (c);
}
T& operator = (const Container& c) {
copy (c);
return *this;
}
T last () const {
return val_ [size_ -1];
}
T val_ [smax_];
int size_;
};
Test Program
int main (int argc, char* argv []) {
Container<int> c;
std::cout << "pushing back 5..." << std::endl;
c.push_back (5);
c.write (std::cout);
std::cout << std::endl;
std::cout << "c.last == " << c.last () << std::endl;
std::cout << "pushing back 3..." << std::endl;
c.push_back (3);
c.write (std::cout);
std::cout << std::endl;
std::cout << "c.last == " << c.last () << std::endl;
std::cout << "pushing back 8..." << std::endl;
c.push_back (8);
c.write (std::cout);
std::cout << std::endl;
std::cout << "c.last == " << c.last () << std::endl;
std::cout << "erasing 3..." << std::endl;
c.remove (3);
c.write (std::cout);
std::cout << std::endl;
std::cout << "c.last == " << c.last () << std::endl;
}
and the results :
pushing back 5...
5
c.last == 5
pushing back 3...
5 3
c.last == 3
pushing back 8...
5 3 8
c.last == 8
erasing 3...
5 8
c.last == 8
if you dont want memory reallocation then you can also use list container i.e linked list ..as it has mostly same properties to the vector..just it do not support random access or []operator ...else vector is perfect:)

Array template class in C++

I can not get the idea of how to create Array template class properly in C++.
The problem is solely out of learning purposes.
Let me provide the code first.
Array.h :
//Developed by Trofimov Yaroslav on 30.03.2018
#ifndef _ARRAY_H_TROFIMOV_
#define _ARRAY_H_TROFIMOV_
#include <string>
template<const size_t n, typename T>
class Array {
static unsigned __freeId, __quantity;
unsigned _id;
T** _array;
const size_t _n;
public:
typedef const bool (* const BooleanResultDelegate)(const T&);
class ArrayError {
const std::string _reason;
const size_t _index;
const size_t _maxIndex;
public:
ArrayError(const size_t index, const size_t maxIndex,const std::string& reason = "")
: _index(index), _maxIndex(maxIndex), _reason(reason) {}
std::string explanation(void) {
std::string res += "Index: " + std::to_string(_index) + "\n";
res += "Max index: " + std::to_string(_maxIndex) + "\n";
res += "Reason: " + _reason + "\n";
return res;
}
};
explicit Array<n, T>(T* arrayFiller = 0)
: _n(n), _array(new T*[n]), _id(++__freeId) {
if(arrayFiller != 0) {
for(size_t i(0); i < length(); ++i) {
_array[i] = new T(*arrayFiller);
}
} else {
for(size_t i(0); i < length(); ++i) {
_array[i] = arrayFiller;
}
}
reportIfDebug<n, T>(*this, "created");
++__quantity;
}
explicit Array<n, T>(const T& arrayFiller)
: _n(n), _array(new T*[n]), _id(++__freeId) {
for(size_t i(0); i < length(); ++i) {
_array[i] = new T(arrayFiller);
}
reportIfDebug<n, T>(*this, "created");
++__quantity;
}
Array<n, T>(const Array<n, T>& that)
: _n(n), _array(new T[n]), _id(++__freeId) {
for(size_t i(0); i < length(); ++i) {
(*this)[i] = new T[that[i]];
}
reportIfDebug<n, T>(*this, "created");
++__quantity;
}
~Array<n, T>(void) {
removeAll();
delete [] _array; _array = 0;
reportIfDebug<n, T>(*this, "deleted", false);
--__quantity;
}
T* operator[](const size_t i) {
if(i > length()) {
throw ArrayError(i, _n, "out of bounds exception");
}
return _array[i];
}
const T* operator[](const size_t i) const {
if(i > length()) {
throw ArrayError(i, _n, "out of bounds exception");
}
return _array[i];
}
const size_t length() const {
return _n;
}
const unsigned getID() const {
return _id;
}
void removeAll(BooleanResultDelegate removeCondition = 0) {
for(size_t i(0); i < length(); ++i) {
if(removeCondition == 0 || removeCondition(*_array[i])) {
delete [] _array[i]; _array[i] = 0;
}
}
}
};
template<const size_t n, typename T>
unsigned Array<n, T>::__freeId(0);
template<const size_t n, typename T>
unsigned Array<n, T>::__quantity(0);
template<const size_t n, typename T>
void reportIfDebug(
const Array<n, T>& instance,
const char* const message,
const bool showContent = true) {
#ifndef NDEBUG
std::cout << "========================================" << std::endl;
std::cout << typeid(instance).name() << ' '
<< message << ' '
<< "id: " << instance.getID() << std::endl;
if(showContent) {
std::cout << instance;
}
std::cout << "========================================" << std::endl;
#endif
}
template<const size_t n, typename T>
std::ostream& operator<<(std::ostream& os, const Array<n, T>& instance) {
for(size_t i(0); i < instance.length(); ++i) {
if(instance[i] == 0) {
os << "[" << i << "]: " << instance[i] << "\n";
} else {
os << "[" << i << "]: " << *instance[i] << "\n";
}
}
return os;
}
#endif
Main.cpp :
//Developed by Trofimov Yaroslav on 30.03.2018
#include <iostream>
#include "Array.h"
int main(void) {
const Array<5, int> a(7);
std::cout << *a[2] << std::endl;
return 0;
}
What is the main problem right now - is that the client of my Array class would have to use [indirection operator *] and [0 value pointer check] to use the objects from array.
I do not want the client to do that. But, if I use reference instead of pointer as a return type of operator[] I will not be able to return types which do not have copy constructor and I will return trash if nothing was put there before.
It seems like in Java and C# the problem is not fixed as well. The user is getting a reference to an object there and definitely should check for null being returned. And [indirection operator *] is called automatically there.

What's wrong with my operator+ of custom Vector class? [closed]

Closed. This question needs debugging details. It is not currently accepting answers.
Edit the question to include desired behavior, a specific problem or error, and the shortest code necessary to reproduce the problem. This will help others answer the question.
Closed 6 years ago.
Improve this question
I'm doing a custom Vector class for education purposes and I can't get the operator+ right. It triggers a breakpoint just after returning a value, but I don't know why, maybe the destructor or constructor, I don't know.
The rare part is, that when I change the operator+ to return T instead of Vector<T>, and returning thiscpy._els[i] the program runs fine. That's why I thought that the issue might be coming from constructor/destructor.
Anyways, here is the relevant part of Vector.h:
#include <initializer_list>
#include <functional>
typedef size_t SIZE;
template <class T>
class Vector {
private:
SIZE _sz;
T *_els;
public:
typedef std::function<void(Vector*)> sorting_function;
static const void populateVector(Vector*, const SIZE&, typename Vector<T>::sorting_function, const bool& sameLength = false);
Vector(SIZE sz = 0) : _sz(sz), _els(nullptr) {}
Vector(SIZE, const T&);
Vector(std::initializer_list<T>);
Vector(const Vector&);
Vector(Vector&& vec) : _sz(vec._sz), _els(vec._els) { delete[] vec._els; }
~Vector() { delete[] _els; }
Vector& operator=(const Vector&);
Vector& operator=(Vector&&);
Vector& operator=(std::initializer_list<T>);
Vector operator+(const Vector&);
SIZE size() const { return _sz; }
T* elems() const { return _els; }
int *begin() { return &_els[0]; } // for (auto i : Vector) {}
int *end() { return &_els[_sz]; } //
};
And here my relevant part Vector.cpp:
#include <stdexcept>
#include "Vector.h"
template <class T>
Vector<T>::Vector(const Vector& vec) {
cout << "Vector initializer" << endl;
if (this != &vec) {
populateVector(this, vec._sz, [&](Vector<T>* obj) {
for (SIZE i = 0; i < vec._sz; i++)
obj->_els[i] = vec._els[i];
});
}
}
template <class T>
Vector<T>& Vector<T>::operator=(const Vector<T>& vec)
{
cout << "Operator = const" << endl;
populateVector(this, vec._sz, [&](Vector<T>* obj) {
for (SIZE i = 0; i < vec._sz; i++)
obj->_els[i] = vec._els[i];
}, true);
return *this;
}
template <class T>
Vector<T>& Vector<T>::operator=(Vector<T>&& vec)
{
cout << "Operator = move" << endl;
populateVector(this, vec._sz, [&](Vector<T>* obj) {
for (SIZE i = 0; i < vec._sz; i++)
obj->_els[i] = vec._els[i];
});
delete[] vec._els;
return *this;
}
template <class T>
Vector<T>& Vector<T>::operator=(std::initializer_list<T> list)
{
populateVector(this, list.size(), [&](Vector<T>* obj) {
SIZE i = 0;
for (T elem : list)
obj->_els[i++] = elem;
});
return *this;
}
template <class T>
Vector<T> Vector<T>::operator+(const Vector<T>& vec)
{
cout << "Operator + const" << endl;
Vector<T> thiscpy(*this);
if (_sz != vec._sz) throw std::runtime_error("Vector size mismatch");
for (SIZE i = 0; i < _sz; i++)
thiscpy._els[i] += vec._els[i];
return thiscpy;
}
template <class T>
const void Vector<T>::populateVector(Vector<T>* obj, const SIZE& newsize, typename Vector<T>::sorting_function repopf, const bool& sameLength = false)
{
cout << "Pupulate vector" << endl;
if (sameLength && (obj->_sz != newsize)) throw std::runtime_error("Incompatible vector length");
obj->_sz = newsize;
try
{
if (obj->_els != nullptr) delete[] obj->_els;
obj->_els = new T[newsize];
repopf(obj);
}
catch (const std::exception& e)
{
obj->_sz = 0;
obj->_els = nullptr;
throw std::runtime_error("Couldn't populate vector");
}
}
And main.cpp:
int main() {
Vector<int> v1{ 1,2,3,4 }; //Vector<T>::Vector(std::initializer_list<T>);
Vector<int> v2{ 2,4,8,16 }; //Vector<T>::Vector(std::initializer_list<T>);
try
{
cout << "----------" << endl;
v1 + v2; //Triggers breakpoint
cout << "----------" << endl;
cout << "done" << endl;
}
catch (const std::exception& e)
{
cout << e.what() << endl;
}
cin.get();
return 0;
}
And the output of the program:
Pupulate vector
Pupulate vector
----------
Operator + const
Vector initializer
Pupulate vector
Your copy constructor never initializes els prior to calling populateVector, so it might not be nullptr (it also might be) and you're calling delete[] on whatever the content of the member pointer is. This can lead to undefined behavior.

STL non-copying wrapper around an existing array?

Is it possible to create an STL-like container, or even just an STL-style iterator, for an existing array of POD-type elements?
For example, suppose I have an array of ints. It would be convenient to be able to call some of the STL functions, such as find_if, count_if, or sort directly on this array.
Non-solution: copying the entire array, or even just references to the elements. The goal is to be very memory- and time-saving while hopefully allowing use of other STL algorithms.
You can call many of the STL algorithms directly on a regular C style array - they were designed for this to work. e.g.,:
int ary[100];
// init ...
std::sort(ary, ary+100); // sorts the array
std::find(ary, ary+100, pred); find some element
I think you'll find that most stuff works just as you would expect.
You can use an inline function template so that you don't have to duplicate the array index
template <typename T, int I>
inline T * array_begin (T (&t)[I])
{
return t;
}
template <typename T, int I>
inline T * array_end (T (&t)[I])
{
return t + I;
}
void foo ()
{
int array[100];
std::find (array_begin (array)
, array_end (array)
, 10);
}
All the STL algorithms use iterators.
A pointer is a valid iterator into an array of objects.
N.B.The end iterator must be one element past the end of the array. Hence the data+5 in the following code.
#include <algorithm>
#include <iostream>
#include <iterator>
int main()
{
int data[] = {4,3,7,5,8};
std::sort(data,data+5);
std::copy(data,data+5,std::ostream_iterator<int>(std::cout,"\t"));
}
You can use Boost.Array to create a C++ array type with STL semantics.
using arrays:
int a[100];
for (int i = 0; i < 100; ++i)
a[i] = 0;
using boost.arrays:
boost::array<int,100> a;
for (boost::array<int,100>::iterator i = a.begin(); i != a.end(); ++i)
*i = 0;
Update: With C++11, you can now use std::array.
A pointer is a valid model of an iterator:
struct Bob
{ int val; };
bool operator<(const Bob& lhs, const Bob& rhs)
{ return lhs.val < rhs.val; }
// let's do a reverse sort
bool pred(const Bob& lhs, const Bob& rhs)
{ return lhs.val > rhs.val; }
bool isBobNumberTwo(const Bob& bob) { return bob.val == 2; }
int main()
{
Bob bobs[4]; // ok, so we have 4 bobs!
const size_t size = sizeof(bobs)/sizeof(Bob);
bobs[0].val = 1; bobs[1].val = 4; bobs[2].val = 2; bobs[3].val = 3;
// sort using std::less<Bob> wich uses operator <
std::sort(bobs, bobs + size);
std::cout << bobs[0].val << std::endl;
std::cout << bobs[1].val << std::endl;
std::cout << bobs[2].val << std::endl;
std::cout << bobs[3].val << std::endl;
// sort using pred
std::sort(bobs, bobs + size, pred);
std::cout << bobs[0].val << std::endl;
std::cout << bobs[1].val << std::endl;
std::cout << bobs[2].val << std::endl;
std::cout << bobs[3].val << std::endl;
//Let's find Bob number 2
Bob* bob = std::find_if(bobs, bobs + size, isBobNumberTwo);
if (bob->val == 2)
std::cout << "Ok, found the right one!\n";
else
std::cout << "Whoops!\n";
return 0;
}