feedback from 9/6/19 reading

Dan reads over diffs from the last virtual forum where sessions PR was discussed. walk through commits since the last F2F forum in May.

[x] Be more explicit about psets only be added and the rank of a given pset doesn't change in the list. Martin raises some issues about "will not decrease" phrase.
[x] MPI_INFO_ENV - do we need to explicitly say its not available in the Session Model? Big discussion of argc and argv. Wesley cited a note from Bill Gropp. Martin wants us to be explicit about MPI_INFO_ENV not being available in Sessions Model. Put in same area where caveats about MPI_COMM_WORLD in the world process model.
[x] check if we need to use \MPI/ inside the mpitermtitlteindexmainsub macro
[x] line 809 in dynamic-2.tex process allocates/instantiates an session handle
[x] for MPI_SESSION_INIT replace during Session instantiation to during this function call.
[x] Capitalize or not Session. Julian (CEA) likes using capital S. Also concerns about needing to qualify as MPI Session vs MPI_T session.
[x] link 855 in MPI_SESSION_INIT prototype remove created
[x] line 874, 911, 912, 920 bare MPI
[x] discussion of first example and how to error handles, remove the assert's after
[x] discussion of MPI_TAG_UB and whether this can vary across communicators allocated from a given session - what should we pick? Discuss with Jim. Also in this area we could change to MPI process
[x] discussion about the length returned from MPI_Session_get_psetlen, keep in sync with MPI_Info related ticket

mpiwg-sessions / sessions-issues