[MITgcm-devel] bottom_ctrl_5x5

[Jean-Michel Campin]

Hi Patrick,

When I try to run the test experiment bottom_ctrl_5x5 with 
some checking option (with ifort, on my laptop),
it stops in S/R mdcg2d_nsa complaining that variable 
"actualits is beeing used without beeing defined"
(I understand, is not assigned to anything).
The original code cg2d_nsa.F initialise actualIts to zero
(and update actualIts to it2d at the end of the it2d loop),
but the mdcg2d_nsa S/R in ad_taf_output.f
skip the initialisation and just assign actualits to it2d 
if/when the residual is larger than the target.
So, this might result in actualits not beeing set.

Do you have an idea why TAF does not keep the initialisation 
of "actualits" in S/R mdcg2d_nsa ?

Thanks,
Jean-Michel