pin_ptr on value struct is needed? [modified]

[Dusan Paulovic]

Hi all, I am working on class to wrapp unmanaged pointer. I am just confused, whether pinning of value struct is needed in C++/CLI, i.e. whether if I will not do it, it can cause some problems. In C#, I am not able to fix structs:

Point pntA = new Point(10, 25);
Point pntB = new Point();

unsafe
{
// Error 4
// You cannot use the fixed statement to take
// the address of an already fixed expression
fixed (Point* ptrA = &pntA, ptrB = &pntB)
{
ptrB->X = ptrA->X;
ptrB->Y = ptrA->Y;
}
}

but, I can simply use address-of operator:

Point pntA = new Point(10, 25);
Point pntB = new Point();

unsafe
{
// Correct
Point* ptrA = &pntA, ptrB = &pntB;
ptrB->X = ptrA->X;
ptrB->Y = ptrA->Y;
}

What is confusing me is that pin_ptr lets me pin structures as well as use address-of operator:

// /clr
generic where T : value struct
T GetValueA(void* ptr)
{
T val = Activator::CreateInstance();

pin_ptr valPtr = &val;
memcpy(valPtr, ptr, sizeof(T));
return val;
}

generic where T : value struct
T GetValueB(void* ptr)
{
T val = Activator::CreateInstance();

memcpy(&val, ptr, sizeof(T));
return val;
}

My questions are: Is pinning needed in C++/CLI functions to fix value struct before using its pointer in unmanaged function? Are value structs allocated in the unmanaged heap? Thanks all, Dusan

modified on Wednesday, August 3, 2011 12:22 PM

[Code o mat]

I think you are asking at the wrong message board, try the Managed C++/CLI[^] one.

> The problem with computers is that they do what you tell them to do and not what you want them to do. < > //TODO: Implement signature here<

[Dusan Paulovic]

Aha, OK, thanks, I have overlooked that one...

[Code o mat]

It happens...

> The problem with computers is that they do what you tell them to do and not what you want them to do. < > //TODO: Implement signature here<