smart pointer release

Hello everyone,


About the release method implementation of smart pointer of COM, there are two approaches below, and approach 1 is preferred is recommended by Inside COM -- should be better.

Anyone know why approach 1 is better than approach 2?

(m_pI is interface pointer to a COM interface of type T, and it is a member variable of the COM smart pointer class)

Approach 1:

Code:

void Release()
{
    if (m_pI != NULL)
    {
        T* pOld = m_pI;
        m_pI = NULL;
        pOld->Release();
    }
}

Approach 2:

Code:

void Release()
{
    if (m_pI != NULL)
    {
        m_pI -> Release();
        m_pI = NULL;
    }
}

thanks in advance,
George

If it's a smart pointer, then what's the point in releasing anything at all? The smart pointer should do that for you; otherwise it's not a smart pointer.
Unless I'm misunderstanding. The both code samples do the same thing, only the first leaves a pOld pointer dangling, which can be dangerous.

Thanks Elysia,


In this case, I am talking about how to implement a smart pointer. Especially, how to implement a smart pointer's release function.

Have I made myself understood?

Originally Posted by Elysia

If it's a smart pointer, then what's the point in releasing anything at all? The smart pointer should do that for you; otherwise it's not a smart pointer.
Unless I'm misunderstanding. The both code samples do the same thing, only the first leaves a pOld pointer dangling, which can be dangerous.

regards,
George

Yes. However, I'd go for the second code. No need for the first.

Thanks Elysia,


I also agree with you. Even if it is original code for Inside COM book, let us just keep it there. :-)

Originally Posted by Elysia

Yes. However, I'd go for the second code. No need for the first.

regards,
George

The reason is probably a misguided attempt to provide some supposed thread safety. Won't work, though, because the things doesn't synchronize in any way.

Approach 1 is used because of exception saftey afaik. Think what would happen if Release throwed.

It's COM. It must not throw in the first place.

I know, I use COM every day at work. Actually I think I've figured it out. It's to do with re-entrancy.

Inside Release, the object's destructor is called if this was the last CComPtr to be holding this object. Now what if for some bizarre reason that object contained a CComPtr to itself, which is what we just called Release on?
First Approach - no problem, during the destructor's call to Release (which is in turn inside the first call to Release) the pointer is already NULL.
Second Approach - Boom! during the destructor's call to Release the refcount gets subtracted again and goes below zero.

Anyway, for both reasons I've stated, Approach 1 is safer.

Thanks CornedBee,


Do you mean there is no need to make synchronization in destructor? Why? Supposing two threads wants to destruct the same object instances at the same time?

Originally Posted by CornedBee

The reason is probably a misguided attempt to provide some supposed thread safety. Won't work, though, because the things doesn't synchronize in any way.

Thanks iMalc!


Your answer is great!

Originally Posted by iMalc

I know, I use COM every day at work. Actually I think I've figured it out. It's to do with re-entrancy.

Inside Release, the object's destructor is called if this was the last CComPtr to be holding this object. Now what if for some bizarre reason that object contained a CComPtr to itself, which is what we just called Release on?
First Approach - no problem, during the destructor's call to Release (which is in turn inside the first call to Release) the pointer is already NULL.
Second Approach - Boom! during the destructor's call to Release the refcount gets subtracted again and goes below zero.

Anyway, for both reasons I've stated, Approach 1 is safer.

regards,
George

Originally Posted by George2

Do you mean there is no need to make synchronization in destructor? Why? Supposing two threads wants to destruct the same object instances at the same time?

Umm no, CornedBee suggested it might be a (rather poor) way of doing thread synchronization, if I understood right.

Thanks Elysia,


My question is, is it possible that two threads are invoking destructor at the same time? If yes, surely we need some synchronization approach, maybe different from the code I posted above. :-)

Originally Posted by Elysia

Umm no, CornedBee suggested it might be a (rather poor) way of doing thread synchronization, if I understood right.

regards,
George

If two threads use the same variable then yes. But since it's a smart pointer, you've probably got a copy of it on another thread, which means it won't call the same destructor.
However, things can still mess up since both can call their respective destructor at the same time. Thread safety is a messy thing. The best thing to do is synchronize access to constructor and destructor, as well as any public access method (and you should do manual synchronize when using the object too).
Otherwise you could run into nasty racing issues.

Originally Posted by Elysia

If two threads use the same variable then yes.

No it is not a problem!

The fundamentals of COM require that incrementing and decrementing the reference count be threadsafe, assuming you've correctly uses CComMultiThreadModel like you're supposed to if your COM object can live in a multi-threaded apartment. In this case it then uses InterlockedDecrement, so only one thread can ever see the refcount be decremented to zero. Furthurmore it would be a bug if the refcount were ever one and multiple threads called Release, though that still would not cause more than one thread to try and delete the object.
Thus only one of the threads can ever call delete on the object.

As I said, I use COM every day at work, and I don't think Elysia has used COM much, if at all.
You can certainly trust Elysia for a quick answer, but often not for a correct one.

Originally Posted by iMalc

No it is not a problem!

Of course it's a problem. It all depends on the implementation of the smart pointer class. Mayhap COM itself may be thread safe, but no the class itself.

As I said, I use COM every day at work, and I don't think Elysia has used COM much, if at all.
You can certainly trust Elysia for a quick answer, but often not for a correct one.

No always correct, no, but no one can be.
I have used COM somewhat, but never really implemented it nor used it in a threaded environment, so that is beyond me.
I don't know the depths of COM, either. I know how to use it, AddRef, Release, QueryInterface, CoCreateInstance, etc. And frankly, COM annoys me, so I avoid it. My knowledge on COM itself isn't perhaps what it should be.