Re: [boost] g++ 2.95.3's STL breaks operator!=

From: Peter Dimov (pdimov_at_[hidden])
Date: 2002-02-09 11:16:30

From: "David Abrahams" <david.abrahams_at_[hidden]>
> From: "Peter Dimov" <pdimov_at_[hidden]>
> > The current ==, != operators compare get(); there wasn't a particularly
> > strong reason to break backward compatibility by making them compare the
> > count.
>
> Considering that shared_ptr<> implicit upcasts now actually *work*, it
seems
> like a good idea to make comparisons work, too. I don't see how it's going
> to break backwards compatibility, anyway. What am I missing?

Hmm.

We now have

template<typename T, typename U> inline bool operator==(shared_ptr<T> const
& a, shared_ptr<U> const & b)
{
    return a.get() == b.get();
}

that works for different T and U, but only when the pointer comparison is
legal.

Are you suggesting that we make operator== work for any unrelated T and U by
comparing the count? It can be done but the NULL needs to be handled
separately fro compatibility.