We document that DerivedExitCode is the highest exit code of the srun
steps, so we need to exclude the batch step. The extern step was already
excluded.

Bug 6196.

Bug 6196

Bug 6196

Bug 6196.

This means we can find batch steps.  NO_VAL isn't used in the code
to return the first one.

Bug 6196

delete_step_record.

so that squeue -s and sstat will show the batch step, and sacct will
show the batch step before the job finishes.

Bug 6196.

It turns out REQUEST_STEP_COMPLETE is never sent by the batch step.

Bug 6196

Run autogen.sh, which picked up one older uncommited change to
a comment line in contribs/perlapi/libslurmdb/Makefile.in.

The correct printf format for size_t is %zu not %lu.

This reverts commit 2707a9c0.

A lot of places rely on a zero size (or, after xcalloc conversion, count)
resulting in a NULL return, so stop xassert()'ing those cases and avoid
jumping to the unpack_error label.

Bug 6430

The code to handle it was already there, this just made it so we
actually parsed it.

Bug 6430

Bug 6436.

Update slurm.spec and slurm.spec-legacy as well.

Update slurm.spec and slurm.spec-legacy as well.

Commit 750cc23e changed the signature for slurm_xmalloc, which
needs to be mirrored here in slurm-perl.h.

Bug 6436.

Adds SPRIO_FORMAT env variable, which is equivalent to the
-o/--format option. Accounted for the new env var in the test suite.

6063

And safe_xmalloc and try_xmalloc as well.

Change slurm_xcalloc() function signature to indicate whether
failure should return NULL or cause a fatal() error. Use that
for these new macros.

Use this instead of handling the multiplication ourselves. This lets us
check for potential problems from integer overflow, and IMNSHO makes the
code look a lot tidier.

Rework xmalloc() and xmalloc_nz() macros in terms of slurm_xcalloc().

Bug 6310.

Update slurm.spec and slurm.spec-legacy as well.

Update slurm.spec and slurm.spec-legacy as well.