COM Question

[Nick Parker]

I have been reading Visual C++.NET by George Shepherd and I ran into something that didn't quite jive with me. Look at the following first:

struct CSpaceship : IMotion. IDisplay
{
ULONG m_cRef;
//...

HRESULT QueryInterface(REFIID riid,
void** ppv);
ULONG AddRef()
{
return InterlockedIncrement(&m_cRef);
}

ULONG Release()
{
ULONG cRef = InterlockedIncrement(&m_cRef);
if(cRef == 0)
{
delete this;
return 0;
}
else
return m_cRef;
}

//...
};

Possibly this isn't a complete implementation of Release(); but shouldn't m_cRef be decremented somewhere in Release();? I can't see how the InterlockedIncrement(); could do this because it is being called in AddRef(); as well as in the Release(); method. -Nick Parker

[valikac]

Interesting. Maybe it it is an error. Otherwise, the server is testing to make sure the client has released all pointers and m_cRef is actually at zero. One situation is multithreading. Kuphryn

[Aravinthan 0]

I think it might be a typo mistake.. i feel it should be interlockeddecrement() function called instead of the increment!! for otherwise.. the component will never die..! or the cref==0 will never happen! as each time the value will always increase! Aravinthan

---

Visit me:
http://www2.domaindlx.com/earavi/
When you know something.. its meant to share with others :-) for otherwise that knowledge has no worth:-)
mail me:
aravinthan@rediffmail.com

---