key_nemocice_decomp still needed?

[smasson]

Context

Do we still need key_nemocice_decomp in v4.0?

Proposal

...

[smasson]

same question for sbcice_cice.F90, still needed?

[acc]

We could generalise this ticket to consider all the cpp keys left in the code. Using a quick grep and sort of the v4.0 code base, I get this set:

key_si3              147
key_agrif            109
key_top               72
key_iomput            63
key_mpp_mpi           42
key_sed               38
key_cice              32
key_vertical          29
key_sed_off           18
key_asminc            18
key_trdmxl_trc        16
key_nemocice_decomp   16
key_floats            15
key_oasis3            10
key_netcdf4            7
key_diainstant         7
key_c1d                7
key_trdtrc             5
key_cice4	       4
key_diahth             3
key_diadct             3
key_cyclone            3
key_nosignedzero       3
key_oa3mct_v3	       2
key_diaharm            2
key_vectopt_loop       1

where the number is a rough count of occurrences. Splitting this into obvious keepers and possible redundancies gives:

Keepers:

key_si3                 
key_agrif               
key_top                 
key_iomput              
key_mpp_mpi     
key_vertical                         
key_asminc              
key_oasis3              
key_nosignedzero 
key_vectopt_loop

Still needed?:

key_cice                                        
key_trdmxl_trc                  
key_nemocice_decomp     
key_floats                              
key_netcdf4                             
key_diainstant                  
key_c1d  
key_sed  
key_sed_off                              
key_trdtrc                              
key_cice4                               
key_diahth                              
key_diadct                              
key_cyclone                             
key_oa3mct_v3                   
key_diaharm

[fdupont]

sorry, missed that one. We are using this key in NEMO3.6 because CICE MPI tiles does not conform exactly to NEMO tiles. There are some solutions to this and CICE6 will finally implement a dynamic allocation. So we will eventually be able to get rid of this. Our goal though is to go directly to NEMO 4.2 + CICE6.