Re: [boost] Question about the weak_ptr constructor with lock() call

Andrey Semashev wrote:

Maybe the cheaper (and arguably more common) version could be used when it is known that the virtual inheritance is not in play?

It's still undefined behavior, even though it would work.