[MITgcm-devel] [MITgcm-cvs] MITgcm/pkg/thsice CVS Commit

[Jean-Michel Campin]

Hi Patrick,

It looks like the changes you made in the argument declaration of
S/R THSICE_SOLVE4TEMP prevent me to compile with option that 
check argument consistency. In:
http://mitgcm.org/testing/results/2013_04/tr_danton_20130423_0/global_ocean.cs32x15/
the error (in make.tr_log_tail):
> mpif77 -m64 -fPIC -convert big_endian -assume byterecl  -O0 -noalign -g -traceback -xHost -warn all -warn nounused -fpe0 -ftz -fp-stack-check -check all -ftrapuv -c thsice_impl_temp.f
> thsice_impl_temp.f(2511): error #7983: The storage extent of the dummy argument exceeds that of the actual argument.   [DTSURF]
>      O          Tice1(siLo,sjLo,bi,bj), Tice2(siLo,sjLo,bi,bj), dTsurf,
> ----------------------------------------------------------------^
> compilation aborted for thsice_impl_temp.f (code 1)
> make[1]: *** [thsice_impl_temp.o] Error 1

I added some comment recently in thsice_solve4temp.F regarding these
declaration:
> #ifndef ALLOW_AUTODIFF
 ...
> #else
> C- note: nothing wrong with above declaration; and below is certainly wrong
> C        when this routine is called from THSICE_IMPL_TEMP
>       _RL flxExSW(1:2)
>       _RL dTsrf1  (1-OLx:sNx+OLx,1-OLy:sNy+OLy)
> #endif
but did not realise it would prevent to compile.

Can I remove these Autodiff specific declaration ?

Cheers,
Jean-Michel

On Sat, Apr 13, 2013 at 04:51:32PM -0400, Patrick Heimbach wrote:
> Update of /u/gcmpack/MITgcm/pkg/thsice
> In directory forge:/tmp/cvs-serv10043
> 
> Modified Files:
> 	thsice_ad_check_lev1_dir.h thsice_ad_check_lev2_dir.h 
> 	thsice_ad_check_lev3_dir.h thsice_ad_check_lev4_dir.h 
> 	thsice_albedo.F thsice_calc_thickn.F thsice_cost_test.F 
> 	thsice_get_bulkf.F thsice_get_exf.F thsice_main.F 
> 	thsice_solve4temp.F thsice_step_temp.F 
> Log Message:
> Prelim. step for adjoint code:
> * change few variable names
> * change iceFlag from logical to _RL
> * exclude some code in case of ALLOW_AUTODIFF (for now)
> * fix (re-instate) various store dirs (but needs double-checking)
> 
> 
> _______________________________________________
> MITgcm-cvs mailing list
> MITgcm-cvs at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-cvs