This is more like a documentation issue than a bug. When making an application side function to receive a handle to an object there is very strange (at first look) way of declaring that function. This is a simple cast function and you use it in the script like this: Obj1@ o1; Obj2@ o2; @o2 = @other_o2; o1 = @Cast2To1(@o2); Here is the engine function registration (the method that works): as_engine->RegisterObjectMethod(class_name, "Obj1@ Cast2To1(Obj2@ ∈ obj2)", asFUNCTION(func_pnt), asCALL_CDECL_OBJFIRST); which look very strange (Obj2@ ∈). Nowhere is documented that you should use that sort of declaration. The application function looks like this: Obj1* Cast2To1(Obj2 *&o2) { return (Obj1*)o2; } I've first tried this: "Obj1@ Cast2To1(Obj2@ obj2)" and received a NULL (zero) pointer in my application function, which was declared like this: Obj1* Cast2To1(Obj2 *o2) { // o2 is always NULL here !!! return (Obj1*)o2; } Don't know why you should declare a reference to a pointer as function argument (when for the Obj1 return handle you use a simple pointer and not *&) but this is the way is working. I think all these kind of issues should be included in the documentation. There is no example of sending a handle parameter to an application function (not even in the feature test sources).

Quote: Original post by WitchLord
Passing an object handle by reference to a function is useful when you want to return more than one value, or if you don't want to have to deal with addref/release in the function.

So basically pasing a reference to a handle is faster since the addref/release functions are not called, right ?

Also, what I've noticed is that there are a lot of addref/release calls when (for me) it seems unnecessary. One example is when you call a method of a class through a handle.

Obj@ h;
@h = @o; // Addref, which is good.
h.FunctionCall(); // Addref on h before call, release after call, BUT WHY ?

Thanks
Licu

[Edited by - Licu on February 2, 2005 3:27:24 AM]

Quote: Original post by WitchLord
The addref before and release after method calls are also due to safety. I have to protect the object from being released before the method is using its attributes. It's quite easy to think of cases where this would cause possible crashes without the extra addref/release calls.

I can figure out a case when this crashes unless you are referring to multithreading. Why a handle refcount will reach zero during my function call ?
I'm talking about an application function not a scripted one.