I just uploaded a new draft of the RTS proposal to the wiki. https://svn.mpi-forum.org/trac/mpi-forum-web/wiki/ft/run_through_stabilizati... https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/wiki/ft/run_through_... Changes: * 2.8: Removed "Similarly, MPI itself provides no mechanisms for handling processor failures." and added the paragraph/advice discussed on the list. * 8.3: Removed modification on the end of the paragraph starting with "After an error is detected," * 13.7: Removed previous update to the Advice to Users based upon clarification in chapter 2. * 17.1: Adjusted last sentence of Advice to Implementors per teleconf. * 17.2: Added definition/citation for 'error' and 'failure' * 17.3: Fixed plural typo in Rationale. * 17.4: Took a pass at clarifying that the process state attribute is consistent to the local process, not all ranks in the application. * 17.5: Fixed a bunch of "rank_states" that should have been "rank_infos" in the function argument declarations. -- Josh -- Joshua Hursey Postdoctoral Research Associate Oak Ridge National Laboratory http://users.nccs.gov/~jjhursey