The safety of returning a new pointer as a reference: MyClass& () { re 1

[Ahnfelt]

My question is simply, will this lead to memory leaks with any otherwise perfect class) or does the compiler know when the newly created variable (created with new but returned as a reference (not a pointer)) will go out of scope? Thanks a lot - Joakim.

#include <iostream>

class Foo
{
public:
  Foo(bool blah) { foo = blah; std::cout << "init" << foo << std::endl; }
  ~Foo() { std::cout << "exit" << foo << std::endl; }
  bool foo;
};

Foo& test() { return *new Foo(false); }

int main()
{
  Foo a(true);
  a = test();
}

/*
  output: (1 is for a true value and 0 is for a false value)
    init1
    init0
    exit0
*/

 

[PerFnurt]

>will this lead to memory leaks

Yes

>does the compiler know when the newly created variable...will go out of scope

No

Foo& test() { return *new Foo(false); }

Returns a new Foo by reference.

a = test();

Assigns a (using the assignment operator) the values of the new:ed Foo. It doesn't transfer ownership of the pointer/reference.

a will be destructed properly then it goes out of scope. But not the new:ed (which is a whole other instance).

Rule of thumb: For every new you must have a delete. Note that you could "hide" and automate the delete by using a std::auto_ptr.

  #include <memory>
  ..
  {
    std::auto_ptr<Foo> foo = new Foo(true);
    ...
    // foo will be deleted when it goes out of scope.
  }

Also, if new fails (as it may do), you should make sure that test() throws an exception as you're not allowed to assign NULL to a reference.

/Per

_{"It was a work of art, flawless, sublime. A triumph equaled only by its monumental failure."}

 

[Ahnfelt]

Thanks for the quick response
I figured it would happen as there was an uneven number of contructions and destructions (see the output: comment), but I had to make sure. I will return a pointer as usual instead