Re: [boost] Re: weak_ptr

From: Peter Dimov (pdimov_at_[hidden])
Date: 2002-09-21 09:58:11

From: "David Abrahams" <dave_at_[hidden]>
> From: "Fernando Cacciola" <fcacciola_at_[hidden]>
>
> > From: "Alexander Terekhov" <terekhov_at_[hidden]>
> > >
> > > Rather: "maybe_dead"{_something}. ;-)
> > >
> > I think it does communicate better the purpose, but it's kind of
> > odd-looking.
>
> I don't like it; it confuses the reference and the referent. THe "weak
> thingie" isn't "maybe dead".
>
> > How about 'untracked' ?, or 'weakly_tracked' / 'weakly_managed' /
> > "weakly_counted"....
>
> There's a long tradition behind the term "weak reference", which I think
is
> a sensible term here. I think an earlier version of weak_ptr had
> dereferencing operations, but now that it does not, "weak_ptr" is a
> less-sensible name.

IMO, weak_reference is not significantly better than weak_ptr to warrant a
name change. I do realize that weak_ptr is not a smart pointer by the "*
and -> canonical definition". On the other hand, weak_ptr is part of the
smart pointers library, it's a _very_ close companion to shared_ptr, and
their implementations are very similar, too. So weak_ptr is good enough for
me, all things considered.