Its just an access violation in Kernel32 once the raw_ function loses scope, which means it could be anything, so I think I'll just drop it. Ah well, I've got to get back to DOS programming now....

Heh, as it reads in the question itself (First Chance exception....), only unless you turn off automatic handling of that exception, it just does a print to the debug window.... Ah well...perhaps I didn't build the object right...

Hi! I'm new to posting here, but I wanted to start out with an annoying feature that was bugging me. I get to do a lot of the weird programming assignments..... I've got a ATL/COM Service that has one Simple ATL Apartment object. It takes a flat file record and converts it into a one row recordset with appropriate columns. It has a method to pass a detached recordset back and forth, such as this: [id(2), helpstring("method ReadRecord")] HRESULT ReadRecord([in, out] VARIANT *iRetVal, [out, retval] IUnknown **ppRecordset); STDMETHODIMP CGenericObj::ReadRecord(VARIANT *iRetVal, IUnknown **ppRecordset) It works perfectly fine, except when you get into the debugger, and only once you get out of the raw_ReadRecord. First-chance exception in Generic.exe (KERNEL32.DLL): 0xE06D7363: Microsoft C++ Exception. But, everything works fine, even in release mode. When I don't detach the recordset, the message does not surface. When the object is passed back in through: STDMETHODIMP CGenericObj::WriteRecord(IUnknown *pRecordset, VARIANT *iRetVal) It does the same thing. The code to generate it: _RecordsetPtr m_pRs; m_pRs.CreateInstance(__uuidof(Recordset)); (read field loop) m_pRs->Fields->Append(_T((LPCSTR)pField.sExternalName), lEnum, lFieldSize, adFldIsNullable); (read field loop end) _variant_t vNull(DISP_E_PARAMNOTFOUND,VT_ERROR); m_pRs->Open(vNull,vNull,adOpenForwardOnly, adLockOptimistic, adCmdUnknown); m_pRs->AddNew(); (read file field loop) m_pRs->Update(); *ppRs = m_pRs.Detach(); Just curious what you coders think....