
  * 2014/4/24 (Kai) -- optional in-tree build of openggcm

	The idea is to build openggcm without running ./runme in a
	separate directory. It's not really that yet, ie., you have to run
	./runme once anyway, but from then on, a simple "make" should redo
	whatever is needed.

	It's off by default, so the first thing is to "./configure
	--enable-openggcm" in target-build/csrc. Then, adapt the runme in
	target-build/csrc/ggcm and run it, and (FIXME) "rm ggcm.f". From
	now on, "make" shall do everything needed to build/rebuild your
	code, including re-running fppn when you change old Fortran code.

  * 2014/4/23 (Kris) -- Coordinate generation can now happen at runtime

      The default is still fortran runme time generation, but to generate
      coordinates at runtime, put something like the following examples in
      your runme:

        ggcm_mhd_crds_gen_type mrc
        crds_gen_x_type   ggcm_x_tanh
        ggx1              $v="-n   $NX     -x0 $XX1     -xn $XX2 "
        ggx2              $v="-x1  -24     -x3  -8.5     -dmm  3    -x5   50 "
        ggx3              $v="-h0  12.0    -hn  100     -hm 3    -hmm 0.140625 "
        ggx4              $v="-b1  0.28125     -b2  0.05   -b3  0.5 "

        # OR

        ggcm_mhd_crds_gen_type mrc
        crds_gen_x_type   ggcm_x_cubic
        crds_gen_x_w0     1.0
        crds_gen_x_w1     150.0
        crds_gen_x_a1     4.0
        crds_gen_x_b1     400.0
        crds_gen_x_w2     2.0
        crds_gen_x_a2     -8.0
        crds_gen_x_b2     -30.0

      To change the y coordinate type, just use "crds_gen_y_type". The
      default for that is ggcm_yz. Check "tests/buildbot_test_checkpoint/runme"
      for a more complete example, and "include/input.defines" for more verbose
      documentation.

      Also note that with runtime coordinates, runme-time coordinate
      generation may not succeed (on trillian for example). This means plots
      of the grid and $RUN.grid, etc. files will not be available. Finally,
      if those runme-time files are created, they may not be exactly the
      same (up to machine precision) as those used at runtime.

	* 2014/4/23 (Doug) -- fix RCM hang and add buildbot tests

Version 3.1

  * 2007/8/29 (JR)  Added option in script.runme to allow for a hand-edited sw data
                    file 'swdata-manual' to be substituted in 'in.$RUN'.
                    Helpful for runs that should be identical except for some
                    SW/IMF data.  Needed this to check if a substorm is
                    triggered by a northward IMF turning.

  * 2007/8/27 (JR)  Added target target.SDSC-bg01 for SDSC BlueGene machine.
                    Note that BlueGene requires the explicit specification
                    of partitions, which may require modifying the target script
                    or editing the batch script.

  * 2007/8/15 (JR)  Modified mhd-corea.for.  Included an extra limiter step
                    'limit3'.  'limit3' forces the flux limiter to be fully
                    diffusive when the velocity exceeds a threshold.  It is
                    hoped that this will prevent 'dtmin' crashes, which occur
                    when the code goes unstable, usually in a non-critical
                    region in the distant tail.  This should kick in very
                    rarely.  Not really tested yet, but seems to have prevented
                    at least one crash.  Does not affect field pusher at this
                    time, which should also be modified to make sure full
                    upwind is selected when speed is too high.


