[cpp-threads] N2052 sequencing revisions & examples (moderately long)

[Jeremy Manson]

Hans Boehm wrote:

> [Slightly deeper]  Can "sequenced before" contributed directly to
> happens-before instead of using a subset for acquire and release operations?
> This would make the model simpler, and bring it closer to the Java model.
> Some of the need for this disappeared.  But I think we still need the
> distinction, since we don't want (all raw operations):
> 
> Thread 1:
> r1 = x; (1)
> y = 1;
> 
> Thread 2:
> r2 = y; (1)
> x = 1;
> 
> to result in a precedes cycle.  But there may be a better way to phrase
> this.  (This is still only a question of presentation, I think.)
> 

Can you promote the subset to the precedes operation?  In other words, 
can you say A precedes B if it A happens before B and A is a release and 
B is an acquire, or B sees A, or transitivity?

					Jeremy