[MITgcm-support] problem running on OS X

[samar khatiwala]

Hi

I've been having trouble running the model on a G4 Mac
(details below).
Its running OS 10.3.4 + absoft compiler and I've never had
problems running the model before. Perhaps someone has a
solution.

Thanks
Samar

Here are the gory details:
I decided to upgrade to c53 and thought I should run one of
the verification expmt to make sure all was well. exp0
compiles just fine (but see below). However, on running the
model I get:
(See http://www.ldeo.columbia.edu/~spk/output for complete stdout.)

...
(PID.TID 0000.0001) SOLUTION IS HEADING OUT OF BOUNDS: tMin,tMax=
-3.211E+34   5.352E+34
(PID.TID 0000.0001) STOPPING CALCULATION
 S/R EEDIE: Only     0 threads have completed,     1 are expected for this
configuration!
 Possibly you have different setenv PARALLEL and nThreads?
 MON_SOLUTION: STOPPED DUE TO EXTREME VALUES OF SOLUTION

Any ideas whats going on?

Another (minor) issue: the makedepend shipped with absoft has
problems with permissions. Apparently, it was written to work
with their IDE and not from the command line. I asked them
about this almost 2 years ago and the said this was a feature
and not a bug. So I use /usr/X11R6/bin/makedepend.
However, by default, genmake2 uses the vendor supplied makedepend.
There are several trivial solutions to this, but it would be
nice if the darwin_absoft build options file had
/usr/X11R6/bin/makedepend hardwired in it.

And another thing: genmake2 checks for the existence of fdate, system,
etc. On my Mac this returns a 'no' for all. However, this is not
right. libU77.a does contain all these routines, so they are in fact
available.