[cpp-threads] Another strawman memory model proposal

[Doug Lea]

> 
>>Unfortunately, I'm also becoming convinced that this is a moot point,
>>since we need to deal with unordered atomics.
> 
> 
> To balance things out, I'm becoming convinced that your original approach of 
> shooting for SC normal form (and relegating the atomics to "formally 
> undefined but we all know what they mean" land, currently occupied by 
> pthreads) may be the only practical approach. 
> 
> 

Or there's the Goldilocks approach that maybe I shouldn't
have given up on -- of only supporting atomics modes we think we
can explain within the SC-ish model, and work on lightening the lightest
of them so people won't complain about performance. I'm still not
sure this goes anywhere either though.

-Doug