> So, are people happier with the idea of having the semantics clearly
> state that these transformations are legal but undesirable from a
> quality of service point of view?
As I said previously in regard to fairness when doing lock coarsening, any
transformation that yields behaviour consistent with some legal
implementation, as per the JLS, is a valid transformation. It then becomes a
QoS matter in my mind. Certainly I think it would be very odd if a VM
actually did "the right thing" for yield/sleep but at the same time the JIT
could transform them into no-ops.
David Holmes
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:37 EDT