[MITgcm-support] upgrade to latest checkpoint

[Christopher L. Wolfe]

Jody,

> My point is that I run the gcm from an "input" directory, and put the
> output in the "results/runA", "results/runB", etc directories.  The
> "input" directory is the launching point for the different results, so
> if "runB" has a different grid than "runA" I've lost the information
> for "runA" if I didn't clean up in time.

What I do in these situations is have an input directory and a bunch  
of results/run?? directories like you, but I just make symbolic links  
from the run directory to the input files. Then I launch the code  
from the output directory. That way I can do multiple runs without  
them stepping on each other's toes, but I don't have to worry about  
the multiple copies of the input files getting out of sync.

Christopher