From: Greg Colvin (gcolvin_at_[hidden])
Date: 2001-06-30 10:26:40
From: Peter Dimov <pdimov_at_[hidden]>
>
> So my point is that a good implementation of thread::ref can directly
> compete with the 'layer 1' noncopyable thread object and be an order of
> magnitude more efficient than 'layer 2' shared_ptr<thread>, while at the
> same time being more user-friendly.
Which is my point too. Thanks for laying it out so clearly.
That said I don't object to providing a two-layer design, but
it is more complicated to do so, and I haven't been convinced
of the need.