[MITgcm-devel] turnoff_model_io ?

[Jean-Michel Campin]

Hi Patrick and others,

I've just changed "pkg/cost/cost_final.F" to switch off useDiagnostics
(it's not different from what is done in ctrl_init_ctrlvar.F and
 ecco_cost_final.F and may be also in few other places):
When I run the verification AD test hs94.1x64x5, I was getting 
funny file "fort.9" because output file dynStDiag.0000000000.txt
was open once (in DIAGSTATS_INI_IO, with unit 9) but was closed 12
times (from DIAGSTATS_CLOSE_IO). 
Now I am no longer getting this "fort.9" file.
It also changes some of the regular diagnostics output files
(I suspect they were over-written several times).

I would not mind having also all the PKG(s)_monitorFreq reset to zero
(in the same way as the main monitor is turned off). It would make 
the standard output smaller, and we might not loose too much.

May be we could add 1 main S/R in model/src (TURNOFF_MODEL_IO or 
something like this) that would take care of this, and could be 
called from different places (may be with a integer argument 
mode in case the selection of what needs to be turned off depends 
on where it's called from).
This way, we would not need to include MNC_PARAMS.h (and others PKG
header files if more flags have to be turned off)
in cost_final.F, ctrl_init_ctrlvar.F, and ecco_cost_final.F (and 
elsewhere ?).
What do you think ?

Cheers,
Jean-Michel