Dear Experts,

Say I have this:

  struct M {
    bool ax;
    bool ay
    bool bx;
    bool by;
  };

I decide I'm going to refactor that to add some hierarchy:

  struct S {
    bool x;
    bool y;
  };

  struct new_M {
    S a;
    S b;
  };

I replace code that used to say ".ax" with ".a.x", etc.

All is good, except that somewhere I take a pointer to a member of M:

void toggle(M* m, bool M::* p)
{
  m->*p = !(m->*p);
}


Both the old M and the new_M actually have the same layout in memory.
This makes me wonder if it's possible to have a "deep" pointer-to-data-member.
Since this is the simple case with no inheritance or virtual methods, the
pointer-to-members are just offsets into the struct (right?). So I feel that
I should be able to create a pointer that refers to new_M.a.x:

bool M::* ptr = &(new_M.a::x);

Is there some flaw that makes this impossible to implement? Or maybe there
is some existing way to do it?

Maybe....

auto ptr = reinterpret_cast<bool M::*>( offsetof(M,a) + offsetof(S,x) );

????


Thoughts anyone?

Thanks, Phil.