[omniORB] Is this possible?

[Sai-Lai Lo]

I'm not sure I understand what you are trying to do. It seems that you are
subverting the type system with pointer manipulations. Something that I
always try to avoid...

>>>>> Bing Zhang writes:

> Maybe this is only a C++ question. I am using VC++ 5.0 on NT 4.0 with
> omniORB 2.4.0.


> In side the function void GIOP_S::HandleRequest(CORBA::Boolean
> byteorder) in file giopServer.cc, there is a piece of code:

>   omniObject *obj = 0;
>   try {
>     obj = omni::locateObject(omniObjectManager::root(),pd_objkey);
>     if (!obj->dispatch(*this,(const char
> *)pd_operation,pd_response_expected))
>         {
>          ......
>   }

> We want to manipulate the pointer obj returned from     obj =
> omni::locateObject(omniObjectManager::root(),pd_objkey); , so I set it
> to a global variable defined as

> omniObject *ptr;

> Inside the eg3_impl.cc main(), a call ((Echo_i *) ptr
> )->echoString("test"); leads to the destructor of omniObject and
> "abnormal termination" is thrown out when it executes assert(0).  I
> found out there is a byte difference between orginal Echo_i pointer and
> the omniObject pointer.

> Echo_i e = new Echo_i();
> cerr << "Orginal ponter address is " <<  e << "The casted omniObject
> pointer address is " << (omniObject *) e << "The global pointer address
> is " << ptr << endl;

> The address of (omniObject *) e and ptr are same, but is different from
> the address of e.

> Am I doing something impossible?

> Bing