slurm_protocol_errno.h provides a default place for slurm specific error definitions until Jim completes his general errno implementation

Cleaned up comm layer , this never ends :)
added pack and unpack routines for new communcations paradigm

Completely reimplemented controller.c top to bottom to accomidate the new comm layer

comm layer method of doing things

these probably need a permanent home some where else.

We should not duplicate definitions in api/slurm.h and slurmctld/slurmctld.h
they should both include a common include file for common definitions

unpacks for client side info calls in progress

  "processes" not "processors").
* fix some dry coding errors in step_mgr.c found when compiling after
  configure --with-elan

Changed a couple of names in the qsw.c api:
 qsw_prog_reap -> qsw_prgdestroy (since it's just a wrapper for rms_prgdestroy)
 qsw_attach -> qsw_setcap (since it's just a wrapper for rms_setcap)
 qsw_signal_job -> qsw_prgsignal (since it's just a wrapper for rms_prgsignal)
Added a conditional include of qsw.h to slurmctld.h so it would compile
when configured --with-elan.

error to malloc/free and printf in api.
Moved api's parse_node_name into slurmctld/node_mgr.c and stopped loading
with libapi.

error to malloc/free and printf.

AIX didn't like a user binding to the loopback,
it threw a errno 68 on bind

Update src/test/runqsw.c to fail if it fails to determine network Id.
Update src/common/qsw.c to work with new driver (elan3_create requires
a handle from elan3_control_open now instead of _elan3_init).

FD_ISSET now returns an int

cleaned up packing and upacking for job_desc
added src/common/log.h include to resolve link warnings

   the only available.

   automake versions.

orientation for greater speed (skips who empty words rather than
checking each bit). - Jette

needed) and WillRun (test if would execute immediately, default is to
actually run it).

lots of cleanup

Changed node_mgr.c to pack config information (rather than actual node
data) iff FAST_SCHEDULE flag is set. The get node info call will return
the data actually used for scheduling purposes, which may not be the
data associated with each individual node.

specified at submit time start with a value of 1<<16 and DPCS uses
values 1 to 0xffffffff. This preclude effective use of a max job_id
in search function, which was able to quickly indicate that a job_id
requested by DPCS was not already in use. I added a job_id hash table
to restore quick job find operations.
Added check of packstr function to insure string would not go over
64k bytes, which is the limit for the protocol used.

specified at submit time start with a value of 1<<16 and DPCS uses
values 1 to 0xffffffff. This preclude effective use of a max job_id
in search function, which was able to quickly indicate that a job_id
requested by DPCS was not already in use. I added a job_id hash table
to restore quick job find operations.

Function was renamed.