Also for the record, appendix A in my PhD thesis has a proof that for both
volatile semantics, the set of programs that is data-race-free is the same.
(Actually, the proof is for even stronger volatile semantics than we are
considering, where any two conflicting synchronization accesses can be used
in the synchronizes-with relation.)
> -----Original Message-----
> From: firstname.lastname@example.org
> [mailto:email@example.com] On Behalf Of Sylvia Else
> Sent: Wednesday, April 07, 2004 7:10 PM
> To: 'javamemorymodel-cs.umd.edu'
> Subject: RE: JavaMemoryModel: Use-case for strong volatile
> At 02:14 AM 3/04/2004 -0600, Sarita Adve wrote:
> >I believe we can prove that the class of programs that are
> data-race-free is
> >identical for the strong and weak semantics.
> If this is the case, then it should be final nail in the
> coffin for the
> strong interpretation.
> a) The model makes few positive guarantees for programs with
> data races
> making it very difficult to predict what the program will do.
> The model is
> primarily about saying what a program with data races will NOT do.
> b) The strong interpretation can complicate distributed
> which compromises their efficiency for data race free programs.
> So where now is the justification for going with the strong
> JavaMemoryModel mailing list -
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:02 EDT