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

[Jean-Michel Campin]

Hi Gael,

You changed the conditions that lead to a stop (as explained in my msg):
> But I have an other issue:
> You did changed exf_check.F much more than this check-in message describe:
> before there was a stop if uwind file or vwind file was not empty
> but ALLOW_ATM_WIND not defined (+ similar check ustress,vstress).
> But now there is a stop if uwind file or vwind file is empty
> but useAtmWind=T (+ similar check for ustress,vstress).
and this is the problem.

I am perfectly fine with doing what you described in the cvs message 
but not with changing the conditions that trigger a stop.

I continue to cc mitgcm-devel since the new conditions you put might
cause problems to some users.

And regarding this:
> Can you revert the logic of these checks as they were before (keeping the
> ALLOW_ATM_WIND <-> useAtmWind substitution, like the check-in message
> description).
If you prefer to let me do it, I will do it.

Cheers,
Jean-Michel

On Sat, Oct 20, 2012 at 11:22:50PM -0400, Gael Forget wrote:
> Hi Jean Michel,
> thanks for the summary fix. If I recall correctly, removing the 
> ALLOW_ATM_WIND condition to the empty file name test, and replacing 
> it with the useAtmWind condition, falls under this part of the cvs message :
> 	ALLOW_ATM_WIND now just sets the useAtmWind default (see
> 	exf_readparms.F) and force defines ALLOW_BULKFORMULAE (EXF_OPTIONS.h).
> I suppose I could have expanded that message further, and mention the 
> checks specifically. The point of the revision was to be able to use the same executable 
> with stress or wind, which I typically needed for the llc90 verification experiment.
> I am not sure how the new check logic is wrong, but it is if I read correctly your message. 
> By the way, I thought you felt like pkg/exf had excessive CPP switches, and should rely
> on such run time switches instead. I thought that this change was a step in that direction.
> Cheers,
> Gael
> 
> 
> 
> 
> 
> 
> 
> 
> 
> On Oct 20, 2012, at 9:33 PM, Jean-Michel Campin wrote:
> 
> > Hi Gael,
> > 
> > I fixed exf_summary.F (printing ustress,vstress parameters) 
> > which was wrong after this Aug 28 modification.
> > 
> > But I have an other issue:
> > You did changed exf_check.F much more than this check-in message describe:
> > before there was a stop if uwind file or vwind file was not empty 
> > but ALLOW_ATM_WIND not defined (+ similar check ustress,vstress).
> > But now there is a stop if uwind file or vwind file is empty
> > but useAtmWind=T (+ similar check for ustress,vstress).
> > 
> > Can you revert the logic of these checks as they were before (keeping the 
> > ALLOW_ATM_WIND <-> useAtmWind substitution, like the check-in message
> > description).
> > 
> > And if you think that these checks need to be changed, please explain
> > why and what needs to be changed (on the devel list) before changing the code.
> > 
> > Cheers,
> > Jean-Michel
> > 
> > On Tue, Aug 28, 2012 at 03:17:46PM -0400, Gael Forget wrote:
> >> Update of /u/gcmpack/MITgcm/pkg/exf
> >> In directory baudelaire:/srv/scratch/gforget/MITgcm/pkg/exf
> >> 
> >> Modified Files:
> >> 	EXF_PARAM.h exf_ad_check_lev1_dir.h exf_ad_dump.F 
> >> 	exf_adjoint_snapshots__g.F exf_adjoint_snapshots_ad.F 
> >> 	exf_bulkformulae.F exf_check.F exf_check_range.F 
> >> 	exf_diagnostics_fill.F exf_getffields.F exf_getforcing.F 
> >> 	exf_getsurfacefluxes.F exf_init.F exf_init_fixed.F 
> >> 	exf_monitor.F exf_monitor_ad.F exf_readparms.F exf_summary.F 
> >> 	exf_wind.F 
> >> Log Message:
> >> 
> >> - pkg/exf : added run time switch useAtmWind to replace ALLOW_ATM_WIND       
> >>  cpp switch. ALLOW_ATM_WIND now just sets the useAtmWind default (see 
> >>  exf_readparms.F) and force defines ALLOW_BULKFORMULAE (EXF_OPTIONS.h).
> >> - pkg/exf, autodiff, ctrl, ecco and seaice : remove ALLOW_ATM_WIND
> >>  brackets, or replace them with useAtmWind ones.
> >> - pkg/ctrl, ecco : allow to compile both ALLOW_U/VSTRESS_CONTROL and 
> >>  ALLOW_U/VWIND_CONTROL. Depending on useAtmWind, one is inactive, 
> >>  and the other is active (see exf_getffields.F/exf_getsurfacefluxes.F).
> >> 
> >> 
> >> 
> >> _______________________________________________
> >> MITgcm-cvs mailing list
> >> MITgcm-cvs at mitgcm.org
> >> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
>