[MITgcm-devel] refining debugLevel choices

[Jean-Michel Campin]

Hi Dimitris,

Thanks for the feedback.

I also propose to move "debugMode" from "data" to "eedata",
so that it will be available for the early initialisation 
phase (usefull when debugging coupler interface). 
I thought about making this move a long time ago, but could be a good
time now to do it, so that the default debugLevel is set 
according to debugMode (and stay unchanged after reading
data 1rst namelist). 
And would add a stop in ini_parms.F if debugMode is (re)set to 
something different from eedata/eeset_parms.F 

Cheers,
Jean-Michel

On Sun, May 22, 2011 at 04:18:35PM -0700, Menemenlis, Dimitris (3248) wrote:
> Jean Michel What you suggest sounds reasonable.
> 
> Dimitris Menemenlis
> 818-625-6498
> 
> On May 21, 2011, at 9:22 AM, "Jean-Michel Campin" <jmc at ocean.mit.edu> wrote:
> 
> > Hi,
> > 
> > I talked about adding a new parameter to refine the choice of
> > messages to print to STDOUT, but when I looked to different printed
> > messages, seems little bit arbitrary which one to switch to condition on 
> > the new parameter and which one to keep with debugLevel. So I changed 
> > my mind and propose instead to extend the choices related to debuLevel:
> > 
> > Would add 2 more levels: debLevC,debLevD= 3,4
> > (presently we have 0,1,2 =debLevZero,debLevA,debLevB)
> > and change the code so that:
> > printed msg with present debuLevel=1  <-> new debuLevel=2
> > printed msg with present debuLevel=2  <-> new debuLevel=4
> > And new debuLevel=1 will be less printed msg than new debuLevel=2 (present debuLevel=1)
> >  (in particular, no MDSIO file opening msg at debuLevel=1)
> > And new debuLevel=3 will be less printed msg than new debuLevel=4 (present debuLevel=2)
> >  (in particular, no debug-stats of all state vars at debuLevel=3)
> > 
> > debugMode=T would switch new debuLevel to 3 (instead of new 4)
> > and the default value for debuLevel would be debLevB ;
> > can also change the default to debLevA for adjoint run (with #ifdef 
> > ALLOW_AUTODIFF_TAMC) to avoid printing messages each time MDSIO open 
> > a file in read mode.
> > 
> > The changes in the code don't look too bad (and if I miss one, 
> > it's not going to break the code) and if one user forgot to
> > make the corresponding changes in his parameter file "data", 
> > it will just result in less printed messages (not too bad).
> > 
> > Do you agree with those changes ? other suggestions ?
> > 
> > Thanks,
> > Jean-Michel
> > 
> > 
> > _______________________________________________
> > MITgcm-devel mailing list
> > MITgcm-devel at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel