[MITgcm-devel] seaice_itd experiment

[Torge Martin]

Hi Jean-Michel,

good points, thanks for the "to do" overview.

1) by introducing the verification experiment Patrick and I discovered that
the pickup is not working properly with ITD (pickup files written with ITD
contain zeros only). That's one issue I am working on, maybe there's more
once this one is fixed ... I'll look for uninitialized variables as well.

2) I grew up with f90, sorry. How do I best check for code not compatible
with g77?

Your help with 1) and 2) would be greatly appreciated.

3) just have removed all "ToM" comments; will remove print statements next;
and check-in on main branch.

Cheers,
Torge


On Thu, May 2, 2013 at 4:51 PM, Jean-Michel Campin <jmc at ocean.mit.edu>wrote:

> Hi Torge,
>
> I have few remarks regarding the seaice_itd verification experiment
> (and related ITD code in pkg/seaice):
> 1) since the new experiment seaice_itd was added, it has not been running
> on many
>  compiler/platform and is not running (in fact, all of the 3 tests) with
> strick
>  compiler checks (the "reference" test on baudelaire, using gfortran and
> "-devel"
>  option, see, e.g.,
>
> http://mitgcm.org/testing/results/2013_05/tr_baudelaire_20130502_4/summary.txt
> ).
>  This is likely due to un-initialised local variable causing a Floating
> exception.
>  It will be useful to fix this, since
>   a) it might also help you.
>   b) once it's fixed, and it run fine, we will be sure that the code is
> safe.
>   c) but it might happen that once this 1rst problem is fixed, with these
> strick-checks,
>     we could discover an other Pb that shows up later in the run/call
> sequence.
> 2) there are few f90 instructions that prevent to compile with g77,
>  this is not very serious, but should be easy to fix.
>  And when it's fixed, the code get tested with more compiler and on
>  more platforms.
>
>  I might be able to help a little bit on these 2 Pb, but I am not sure
>  if the current pkg/seaice version is "up to date" - since you checked in
> some
>  other thing in contrib.
>  This is one reason why it would be better to make future changes in
>  pkg/seaice instead of in contrib, so that it's tested, and everyone works
>  with tested version.
>
> And regrading the ITD code, just 2 details:
> I don't think it's really useful to keep commented lines like:
> > CToM<<<
> > C>>>ToM
>  since we can always check with CVS (and cvsview) who made which
> modification
>  (+ it's generally within #ifdef SEAICE_ITD / #endif).
>
> And also, it would be nice to cleanup the many debug prints in
>  pkg/seaice/seaice_model.F as mentioned earlier
>  http://mitgcm.org/pipermail/mitgcm-devel/2012-October/005642.html
>
> Cheers,
> Jean-Michel
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20130503/e907cbbe/attachment.htm>