<html><head><meta http-equiv="Content-Type" content="text/html charset=windows-1252"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div>Here's the latest version of the slides from yesterday's call. This now includes:</div><div><br></div><div>* Misc. small corrections throughout</div><div>* [8] An alternate slide for failure notification if we decide to combine failure_ack & get_acked to reduce confusion.</div><div>* [10] Reworked rationale for MPI_Comm_revoke to reduce confusion about a possible (though not used) alternative. I agree that this slide is still weak. It may be good enough to just say that we can implement this functionality on top of MPI and leave it at that. Then we could just remove this slide entirely.</div><div>* [13] Corrected code describing how to validate a communicator after calling a creation function</div><div>* [16] New slide describing one-sided semantics including the state of memory after a failure.</div><div>* [17] New slide describing how to handle passive target locks after a failure.</div><div>* [18] New slide describing file I/O semantics.</div><div>* [20 - 27] New slides with extended library construction example using ScaLAPACK.</div><div>* [28] New slide describing implementation status</div><div><br></div><div>More comments are always welcome.</div><div><br></div><div><div>Thanks,</div><div>Wesley</div></div><div><br></div></body></html>