Traditional Singleton Implementation vs Static Object Singleton Implementation

I have usually implemented Singletons as described in the Design Patterns book. Recently someone suggested another implementation of Singleton using creating a static instance of a class and returning that instead of using new and returning a pointer to the new singleton. I would like to get your input. Is there any advantages of using one or the other? Disadvantages? Disadvantages for the static singleton implementation?

Let me illustrate:

Traditional Singleton Implementation

Code:

---

Car* Car::getInstance()
{
  if(!InstanceFlag)
  {
      Instance = new Car();
          InstanceFlag = true;
          return Instance;
  }               
  else
  {
          return mp_Instance;
  }
}

---

Static Implementation

Code:

---

Car& Car::getInstance()
{
  static Car theCar;
  return theCar;
}

---

Advantages & Disadvantages? Any documents or articles I can refer to for the second implementation ? Thanks for your help!

Looking only at provided code, I can say that:
1. First version is not thread-safe.
2. In the second version Car instance is created even if getInstance is never called. Not optimal, especially if the car is expensive...

#2 is also not thread-safe in C++03. It is in C++0x

In #2, storage space is pre-allocated for Car, but the constructor is only called at the first call to getInstance().

gg

Quote:

---

Originally Posted by Codeplug

#2 is also not thread-safe in C++03. It is in C++0x

---

Is there documentation on this? I hadn't heard it before.

Quote:

---

Originally Posted by Lindley

Is there documentation on this? I hadn't heard it before.

---

the draft in my possession reads

Quote:

---

Originally Posted by 6.7.4 Declaration statement

The zero-initialization (8.5) of all block-scope variables with static storage duration (3.7.1) or [...] Otherwise such a variable is initialized the first time control passes through its declaration; such a variable is considered initialized upon the completion of its initialization [...] If control enters the declaration concurrently while the variable is being initialized, the concurrent execution shall wait for completion of the initialization.

---

Alex, what do you mean in #2 an instance of the Car is created even if getInstance is not called? Can you explain?

Codeplug, for #2 memory is preallocated and if you were running out of memory you would know at build time instead of run time like with dynamic memory allocation I ran into issue trashing the heap. In that case #2 might be better.

Lindley, Im have not found documentation on this way of implementing a Singleton (#2) and that is why I decided to post this topic.

Thank you all for replying. I'm still trying to figure it out whey #1 may be better than #2 or viceversa. Keep in mind that #1 is the solution suggested by the Design Patterns book and what I have seen in the industry. Thanks!