Re: [boost] Usefulness of weak_ptr

From: Darin Adler (darin_at_[hidden])
Date: 2002-02-12 12:42:52

On 2/12/02 9:02 AM, "Peter Dimov" <pdimov_at_[hidden]> wrote:

> Done, see the CVS. Comments appreciated. We probably need to decide what to
> do with the *_nmt variants so we can remove atomic_count (it's no longer
> needed.)

I think that bad_weak_to_shared_cast should be part of the public interface,
not part of boost::detail, since it can be thrown by the shared_ptr
constructor that takes a weak_ptr parameter.

I think we should remove the "// may throw" comments, because they are more
likely to be mistaken for "// never throws" than to be helpful.

I'm not certain that retaining operator-> for weak_ptr is a good idea,
because the lifetime of the implicit shared_ptr might be non-obvious.

I'll wait a few days for this to settle down before updating the
documentation, unless you plan to do it.

    -- Darin