for a job as well as the nodes to be allocated and processor count
    (for use by Moab).

Add job and task memory limits to job step credential
Add enforcement of job and task memory limits to job accounting and 
  abort job that exceeds the limit

    sbatch commands.

    accounting data collection polling interval.

Added srun option --checkpoint-path for use with XLCH (Hongjia Cao, NUDT).

    files are opened ("t" for truncate, "a" for append).

SRUN_PING, SRUN_USER_MSG, and SRUN_EXEC (including 
OpenMPI checkpoint).

    job steps to be queued when resources are not available within an existing
    job allocation to dedicate the resources to the job step. Useful for
    executing simultaneous job steps. Provides resource management both at
    the level of jobs and job steps.

    to task 0 (all other tasks have I/O discarded)

with Munge an alternative.

That functionality is now available in salloc, sbatch, and sattach,
respectively.

Also removed the forked message process from srun.  The slurmctld will no
longer kill a job allocation just because it cannot ping srun, so it
is no longer critical to keep a message process running when the user
stops the srun process.

code. Applies only to non-batch jobs. and the bad exit code was only 
reported by "scontrol show jobs" (the exit code in accounting data 
was correct).

MPI plugins no long access structures in either srun or the slurmstepd.

Also, MPI plugins allocate state structures for each job step instance.

This allows the MPI plugins to be used in the SLURM API slurm_step_launch()
function.  It also allows slurm_step_launch() to be called multiple times
from within the same process.

the path to srun as determined by argv[0].