> In those cases I would prefere an optimization hint saying
> that it is enough that a put becomes visible to get operations
> only after a call to WINDOW_IN rather than restricting MPI
> 1-sided to the "delayed consistency" model.
We reached consensus that hints do not change MPI semantics. This particular one
does and therefore it is a directive. I can easily write a piece of 1-sided code
following current semantics that works just fine but deadlocks if you aply this
directive. It would not happened if it was a hint.