[boost] Re: Re: Intrusive smart pointer

From: David B. Held (dheld_at_[hidden])
Date: 2002-04-25 12:16:25

"Andrei Alexandrescu" <andrewalex_at_[hidden]> wrote in message
news:aa87k8$k4r$1_at_main.gmane.org...
> [...]
> I'd argue in favor of less proliferation of all sorts of smart
> pointers, for obvious vested interest reasons :o).

What do you think of Dietmar's suggestion that there should be a single
common smart_ptr<T> interface (with different common policy variations,
I suppose?), for interoperability reasons, and a separate policy-based
smart pointer that is possibility the implementation(s) of the simple
interface?  I guess what I don't understand about Dietmar's argument is
just how many "simple" smart pointer variations his interface should
provide?  Is the current <scoped | shared><pointer | array><separate |
intrusive> sufficient, where everyone just pays for dereference checking
and multithreading support?  Or should there also be a set of <single-
threaded | multi-threaded> variants as well?  Only Dietmar can answer
that, I guess.

Dave