com

[avsramesh]

What is the advantage of using CComPtr class. If I use it,is it required for me to call the Release() function of it.
Ex:
CComPtr <IAdd> pAdd;
pAdd->Addition();

pAdd.Release();// Is this really required or not.

[claudiu_t]

Hi,

The answer is NO. As you already have read in the doc the CComPtr class is a smart pointer class, whose purpose is - besides others - to remove the necessity of Release() call.

Now, if you can write a simple console app such as:

void main() {
    ::CoInitialize(NULL);

    CComPtr<ISomeInterface*> ptr = NULL;
    ptr->CreateInstance(CLSID_Library);

    // consume ptr

    ::CoUninitialize();
}

you may notice an error message at run-time! if you add the line

    ptr->Release();

just before the ::CoUninitialize() call the error message disappears. Oops! it seems our smart pointer class is not working properly. In fact is just a variable scope issue: our ptr variable live after the ::CoUninitialize() call and his destructor is called only after this, which is perfectly correct. The workarround is to place the main code in a block, like this:

void main() {

   ::CoInitialize(NULL);

   {
       CComPtr<ISomeInterface*> ptr = NULL;
       // rest of the code here
   }

   ::CoUninitialize();
}

which obviously makes useless the ptr->Release() call.

HTH regards,
Claudiu