C++ Placement New: Tracking down a crash

See my previous two posts on an introduction to placement new if you are unfamiliar with the subject.

• https://blogs.msdn.com/jaredpar/archive/2007/10/16/c-new-operator-and-placement-new.aspx
• https://blogs.msdn.com/jaredpar/archive/2007/10/17/c-placement-new-and-allocators.aspx

Recently I did a bit of work on the heap management story for our code base.  Mainly it was a change to unify the different ways we accessed our internal heap.  In the process I unified our operator new story.  Part if this involved unifying our placement new allocation overloads.  We have several allocators in our code base and I noticed we had several overloads which essentially did the same operation

 static inline
void * _cdecl operator new(size_t cbSize, MyCustomAllocator &allocator)
{
    return allocator.Alloc(cbSize);
}

static inline
void * _cdecl operator new(size_t cbSize, MyCustomAllocator *allocator)
{
    return allocator->Alloc(cbSize);
}

This seemed a bit redundant to me so I deleted the one which contained a pointer overload.  A while later I tested my changes and our application almost immediately crashed.  It only took a few minutes to discover the problem.  We had a lot of operations that followed this pattern. 

 Student *p = new (pAllocator) Student();  // pAllocator typed to MyCustomAllocator*

As I said in my previous posts, new is just another C++ function.  As such it participates in overload resolution.  This code now binds to the placement new operator and not the placement allocation overload which takes a reference.  At first I thought about using a regex search and replace to solve the issue.  However I decided this wasn't a complete solution.  There is no guarantee that my regex will catch every case and any case I miss won't crash until we actually hit the line of code. 

The best case scenario here is to turn that line into a compile error.  That would guarantee that I fix every location that is a problem.  Since new participates in overload resolution you can solve this with a template. 

 template <typename T>
static inline
void* _cdecl operator new(size_t cbSize, T* allocator)
{
    allocator->ThisWillForceACompileError();
}        

This caused all places where a pointer type was passed to new which wasn't explicitly "void*" to turn into a compiler error.  This quickly outlined all of the places I needed to fix up and guaranteed that my fix didn't allow developers who were accustomed to using a pointer to the allocator to accidentally introduce a bug into the code base (myself included :)).