[Mitgcm-support] ini_garrays

[mitgcm-support at dev.mitgcm.org]

Hello,

I tried to see why -fast cause Pb on xolas, but did not find
anything conclusive.
(just in mom_calc_hfacz.F _RS hFacZOpen is better than _RL hFacZOpen)

But still something strange in ini_fields :
ini_garrays is never called (I checked putting a stop in it)
and this might cause Pb on machines that do not initialise arrays.

I changed this (not yet checked-in).
I also set to zero uVel,vVel,wVel,theta,Salt,etaN (in ini_garrays), so that 
whatever the exchanges we use, the overlap region of all Dynamics arrays 
(DYNVARS.h) will be initialised.
Think about changing the name of ini_garrays.F to ini_dynvars.F 
or "set_fields_tozero.F" ?

See you,

Jean-Michel

PS: Concerning the low Eta field, I didn't look further to the results of 
the previous run (AIM on C.Sph, don't remenber where all the results are) 
will see on monday.