C++ Debugging "smell" -

I was debugging a strange problem where an object VMT suddenly points to the base object's methods.

  class base {virtual zero foo () {}} class derived: public base {void foo () {}} derived derived * d = new; ... very complex fettling ... d-foo (); // help! Base: foo () is called !!!   
 This was a bug effectively in the 'complex' code  deleted;  I will not go into details except to tell that  boost :: noncopyable  is your friend, and never under any circumstances does not ever roll their smart pointers classes.  
 But my question is - does this kind of change of VMT have a good 'smell' that you are working with a deleted object? I think VMT 'gets' unrolled' on ' base  during destruction'?  
 Obviously this is very dependent on implantation, and I was 'lucky' that the memory was not crushed on ...   
 
 
 Definitely what the question is ??? But IA says that this:  
 
 I was 'lucky' that the memory was not crushed ...   
 is wrong: you lucky is ???? You were lucky  united  were removed during the time of memory being crushed, then you probably would have a code accident near the real problem and the error would be very clear.  
 The code fails as soon as possible, promptly about the failure location as soon as possible.