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

[Jean-Michel Campin]

Hi Dimitris,

My impression is that having a warning instead of error+stop
is a good compromise (and also match the check-in msg description :-) ):
if, a user, in the process of updating his code,
leaves #define SEAICE_MULTICATEGORY in SEAICE_OPTIONS.h,
which does not do anything anymore, since it has been removed from the code,
there will be a clear warning message - but it offers the advantage of
not beeing force to update it at the same time, and use the same 
customized code dir for the new and old version).
But it's just my personal impression.
That beeing said, in seaice_check.F, we find both type
of messages for retired CPP options (SEAICE_EXTERNAL_FORCING
gives an error but does not stop - would be better to have a
warning there - and SEAICE_ALLOW_TD_IF gives an error and stops).

Updating SEAICE_OPTIONS.h :
1) should definitively update pkg/seaice/SEAICE_OPTIONS.h
2) can be tricky to update all verification/*/code*/SEAICE_OPTIONS.h
 since some have not been updated for a long time, and, in addition,
 there is SEAICE_GROWTH_LEGACY that will be removed soon. 
 So, I would prefer to see only the "almost" up-to-date beeing updated
 now (as you did previously): 
   1D_ocean_ice_column/code/SEAICE_OPTIONS.h
   offline_exf_seaice/code/SEAICE_OPTIONS.h
   seaice_obcs/code/SEAICE_OPTIONS.h
 and postpone the update of the others until SEAICE_GROWTH_LEGACY is gone.

Cheers,
Jean-Michel

On Fri, Dec 21, 2012 at 07:22:43AM -0800, Dimitris Menemenlis wrote:
> Jean-Michel, my preference would be not to leave retired CPP options
> in the verification experiments.  It's confusing when you have not
> followed CVS check ins for a while and then try to update an
> experiment.  It's also confusing for MITgcm users who rely on
> verification experiments as initial templates for their work.
> 
> If you agree with above, I can remove remaining SEAICE_MULTICATEGORY
> from the verification experiments later today.  I promise to be more
> careful and to run the full suite of experiments before check in.
> 
> If you prefer to leave as is and change "Error+Stop" into a
> "Warning", that's fine too.  Let me know.
> 
> Dimitris Menemenlis
> 
> On 12/21/2012 06:50 AM, Jean-Michel Campin wrote:
> >Hi Dimitris,
> >
> >From this cvs-commit message, I thought it was a just warning.
> >But when I look to the code (and check the experiments that are
> >now failing), it's an Error + Stop.
> >
> >So, here is my question:
> >- should we transform this Error+Stop into a warning ?
> >or
> >- should we keep this Error+Stop and make the necessary changes to
> >the few broken experiments (there is also the baltic experiment test
> >run by Oliver on beagle, e.g.:
> >http://mitgcm.org/testing/results/2012_12/tr_beagle-ifort_20121221_2/summary.txt
> >which are affected) ?
> >
> >Cheers,
> >Jean-Michel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel