more fixes for PR 55

[hppritcha]

• [x] simplify first sessions example to be like library_init
• [x] redo figure for sessions to not have wavy lines and to have correct
• [x] review more places in standard with MPI_COMM_SELF and adjust as needed

And more todos from 12/5/18 Forum reading feedback:

• [x] Martin wants to drop “unless the application explicitly calls MPI_INIT” in page 416, lines 46-47. Add valid as an adjective to MPI COMM WORLD in this section.
• [x] We need to fix the second paragraph of 10.1 to better integrate the dynamic process model back into the 3 models for process creation
• [x] fix 10.1 subtitle!
• [x] page 417 line 22 add Window or File
• [x] water down line 45 on page 416 or remove
• [x] question about MPC-MPI MPI process is a thread model from Aurelian and Martin. We need to think about this – maybe add a blurb in the pset definition
• [x] Maybe move last paragraph in 10.4 up to intro part of chapter and reword a bit. remove backward compatible?
• [x] swap 10.4.1 with 10.4.2.
• [x] page 418 line 1 remove use of “runtime” in that first, todo: decide if we want to mandate or highly recommend mpi://UNIVERSE and mpi://comm type shared.no examples.
• [x] make clearer what’s going on with lines 4-5, what exactly is a pset. define semantically?
• [x] key is mandatory for all process sets - Jeff Squyres thinks this needs more verbiage.
• [x] add some more discussion about the flags bit mask to MPI Session init.
• [x] fix f08 binding for MPI_Session_init flags should be TYPE(MPI_Flags), check other f08 bindings too where INTEGER should be replaced with a TYPE
• [x] could we use info arg to MPI_Session_init rather than flags?
• [x] use tools cvar/pvar approach for psets and their lengths
• [x] replace MPI_Errhandler with MPI_Session_errhandler in multiple places (actually this is wrong)
• [x] add text for thread level support
• [x] Aurelian asks about thread safety for things like data type manipulation, etc. This needs to go with the info ops before MPI Init discussion. - [ ] an advice to users about sessions and how implementations may use them – Dan talks about isolation and multiplexing, grouping of communicators. (decided this is a user manual item)
• [x] should we call Sessions Enclaves, Instances?
• [x] pre-defined info keys
• [x] need to add local to the description of MPI_Session_init
• [x] be consistent with capitalization of session/enclave Martin says put MPI in front of Enclave or Session
• [x] function name for MPI_Session_init line 46 on page 418
• [x] make session handle in out to MPI_Session_finalize
• [x] need a lot more text for MPI_Session_finalize – use analogous not equivalent, should we add a drop-all-resources type of call that frees all resources associated with a session. Collective. Files, windows, communicators, groups get leaned up
• [x] MPI Session get num sets replace pset_name with npsets or something like this, same issue with get_psetlen
  - [ ] could we replace all this with an in/out info object
• [x] definitely need to keep the figure
• [x] rework chapter describing MPI_SESSION_GET_NTH_PSET about room for NULL terminator etc. Check MPI T and info in the code example title add a reference to chapter six functions.
• [x] question about where the example would block
• [x] verbiage about session finalize associated with the example
• [x] uniqueness of uri (Jeff says don’t use this!!!) across sets of processes, suggestion and advice to users for MPI_Create_comm_from_group gov://llnl thingy
• [x] should we rename MPI_GROUP_FROM_SESSION_PSET . maybe return MPI GROUP NULL if not valid pset name?

More todos/feedback from 1/30/19 virtual forum:

• [x] unless you call MPI init MPI_COMM_WORLD is not valid (add to the world process model section of chapter 10)
• [x] an instead of a in several places
• [x] line 1101 in dynamic-2.tex INITIALIZE replace with INIT
• [x] scope of unique set name - Martin asks about this for process sets. mpi://self problem. disjoint and unique - maybe add a picture? Dan gives example of MPI_COMM_SPLIT with each process providing a different key - [ ] MPI_SESSION_GET_NTH_PSET - may need something different for Fortran (check with Nathan). Martin wonders if this will cause a problem for PMPI. tool wanting to add its own process set, example of tool generating the PSET NAME. maybe already add NULL terminator even for Fortran.
• [x] check for MPI_SESSION_GET_INFO for session type
• [x] add FORTRAN seesions example

[hppritcha]

notes from 2/25/19 webex:

more discussion of replacing the bit flags mask with info key/values. Pretty convinced we can go with info and drop bit flags. Leave as is for the reading week of 3/4.

functions callable before MPI_Init, MPI_Session_init, etc. is deferred to the work being done by FT wg. MPI_Info_create and related accessors and destructors, in addition to MPI errorhandler constructors and destructors.

discussion of changing from session to something different. Maybe take a vote at the 3/4 reading.

discuss uri arg to MPI_Comm_from_group. Need to find out from Jeff his objections to uri format. We should probably say a little more about the uniqueness of uri.

discuss MPI_Group_from_pset - different name? also return MPI_GROUP_EMPTY if pset name is not valid.

Should double check the Fortran.

need to sync up with tools people about MPI_T and sessions. Should have something done in time for June '19 forum. maybe as errata for 4.0, or delay till 4.1?

[dholmes-epcc-ed-ac-uk]

All items ticked off - either completed or copied to another issue.