From: williamkempf_at_[hidden]
Date: 2001-07-02 14:30:16
--- In boost_at_y..., "Peter Dimov" <pdimov_at_m...> wrote:
> From: <williamkempf_at_h...>
>
> > No, I don't think it illustrates that at all. At best it
illustrates
> > that Boost.Threads should have a thread_group concept to insure
> > other's don't make this mistake in their own implementations.
>
> So to avoid providing a thread::ref (a custom smart pointer) you're
going to
> provide a custom container instead - a higher level component.
Yes. It's more efficient. The ref-counted concept still exists, so
this is a smarter design. It's in line with current C++ design
(iostreams), it's efficient, and it's flexible when used with other
concepts.
> > It does not argue for a ref-counted design in any way.
>
> What would?
Either a usage pattern that's not possible with the noncopyable
design or evidence that performance is not going to be impacted by
ref-counting. Again, look at this from the design criteria of the
C++ language. You should pay for only what you use.
Bill Kempf