[Mitgcm-support] Re: packages_init_....

[mitgcm-support at dev.mitgcm.org]

I had to create them to fit initialization of
OBCs around initialize_fixed and initialize_varia
OBCS affects ini_cg2d (called from init_fixed) and
also has to modify the initial conditions (called
from init_varia).

If we had package_init without the _fixed and _varia suffix
where shuold package_init get called from? initialization_fixed
or initialization_varia?

I personally can't abide the _fixed _varia bits 
but it's needed for (data assim.) check-pointing.
My understanding is that arrays that change with
time (GM/KPP/OBCS) need to be init'd in _varia.

A.

chris wrote:
>  Can anyone explain what we mean with the new
>  packages_init_fixed and packages_init_variables
>  routines? The packages setup is going to be the main way
>  many people might contribute to the core code for at least
>  the next year or so. While the innards of a package and the
>  innards of the core model can be a bit fuzzy, the interface
>  between a package and the core model needs to be pretty
>  clear and well defined.
> 
>  The note Alistair and I wrote on this setup last summer did
>  not have a .._init_fixed because nobody could agree on
>  how to define it. Now it has crept in with the obcs package.
>  Is it needed - what is a clear definition?
> 
> Chris