[MITgcm-devel] halfpipe_streamice verification expt

[Jean-Michel Campin]

Hi Dan,

To remove a file from the CVS repository (technically, the file will not be 
lost but will be moved to the 'Attic" dir) is easy:
1) checkout the code 
2) cd to the dir where the file_to_remove is staying
3) rm file_to_remove (or: mv file_to_remove file_to_remove.rm)
4) cvs remove file_to_remove
5) cvs commit file_to_remove

And just for the update, the experiment halfpipe_streamice is now (last night) 
 running fine in all the open64 & patchscale & pgi testing.
Also, I sometime run some ifort testreport on my laptop that would also 
pickup the problem that Martin's uv100 test shows (check for consistent
argument list).

Cheers,
Jean-Michel

On Fri, Jun 14, 2013 at 07:11:32PM -0400, Daniel Goldberg wrote:
> Hi j-m -
> 
> Quick response on 3 - actually that is an old s/r and WHOOPS it should not have been included in the pkg folder. I don't understand how it is being called, but I will look into it. Is it as difficult to remove files from cvs as directories?
> 
> Sent from my iPhone
> 
> On Jun 14, 2013, at 6:46 PM, Jean-Michel Campin <jmc at ocean.mit.edu> wrote:
> 
> > Hi Dan,
> > 
> > I changed halfpipe_streamice/input/data.streamice, and 
> > might be enough to run with open64 and, hopefully, also with pathscale.
> > Will see how this night tests go.
> > 
> > Regarding the other tests that failed,
> > 1) we can drop g77 (fortran 90 instructions) 
> > 2) I added a genmake_local file to the build dir, hoping it will fix
> >   pgi test on old aces cluster:
> >  http://mitgcm.org/testing/results/2013_06/tr_aces-pgi_20130614_0/halfpipe_streamice/make.tr_log_tail
> >  which does not like lines like: 
> >      IF (X<-1.0*W/2.0) THEN
> >   in eta_gl_streamice.F
> > 3) And also issues with Martin's test on uv100:
> > http://mitgcm.org/testing/results/2013_06/tr_uv100_20130614_0/halfpipe_streamice/make.tr_log_tail
> > -> SLOPE_LIMITER function called with 2 arguments in streamice_advect_thickness_x.F
> > (but this function has only 1 argument !)
> > 
> > Cheers,
> > Jean-Michel