tr1::shared_ptr and v3's atomic ops

[Paolo Carlini]

Jonathan Wakely wrote:

>1) could compare_and_swap be resurrected, or is the porting headache too
>much bother?
>  
>
In principle, I don't see serious difficulties with this. From a
practical point of
view, however, I'm pretty sure it cannot be implemented on some older archs,
like i386, which we are still interested in supporting. Therefore, we should
envisage two different possibilities (lock-free/not lock-free) selected
at config
time. Maybe, in the 4.0 timeframe, better initially providing a locked
baseline.

Paolo.