For this recipe, we require three files. The first is the implementation that we wish to test (we can call the file leaky_implementation.cpp):

#include "leaky_implementation.hpp"

int do_some_work() {

  // we allocate an array
  double *my_array = new double[1000];

  // do some work
  // ...

  // we forget to deallocate it
  // delete[] my_array;

  return 0;
}

We also need the corresponding header file (leaky_implementation.hpp):

#pragma once

int do_some_work();

And, we need the test file (test.cpp):

#include "leaky_implementation.hpp"

int main() {
  int return_code = do_some_work();

  return return_code;
}

We expect the test to pass, since the return_code is hardcoded to 0. However, we also hope to detect a memory leak, since we forgot to de-allocate my_array.