smart pointer release

[George2]

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:

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

Approach 2:

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

thanks in advance,
George

[weaknessforcats]

I have no idea since the destructor of CComPtr uses Approach 2.

[George2]

Hi weaknessforcats,


Any comments?

--------------------
Imagine this situation: object A holds a smart pointer to object B,
which in turn holds a smart pointer to object A. For whatever reason, A
decides to release its reference on B, by calling Release() on the smart
pointer. This happens to be the last reference, so B destroys itself. In
its destructor, it releases its reference on A - again by calling
Release(). This, too, happens to be the last reference on A (the two
objects kept each other alive with a circular reference), so in its
destructor A calls Release() on its smart pointer to B.

The bottom line of this scenario is that smart pointer's Release() ends
up being called from itself, recursively. Approach 1 handles this
gracefully, but Approach 2 will enter an infinite recursion.
--------------------

I have no idea since the destructor of CComPtr uses Approach 2.

regards,
George