cni-changes documented?

[Tom Tromey]

>>>>> "Hans" == Boehm, Hans <hans.boehm@hp.com> writes:

Hans> (I'm not sure whether the smart pointers to implement
Hans> null-pointer checks are implemented as C++ source.)

Hans> At least in my experience, a general problem with C++ "smart pointers"
Hans> is that they only behave 95% like pointers, and it's usually still
Hans> possible to get premature deallocation if you try, or if you try to
Hans> manually optimize too aggressively.

I think our plan for smart pointer is just a very thin wrapper around
the pointer to do null checking:

template<typename T>
class ref
{
private:
   T *val;
public:
   ... constructors and stuff...

   T *operator () const
   {
#ifdef NO_SEGV_HANDLER
     if (! val) throw new NullPointerException();
#endif
     return val;
   }
};


The hope is that the overhead from this approach will be very small.

Tom