[MITgcm-devel] optfiles cleaning

[Jean-Michel Campin]

Hi Dimitris,

I have a question regarding optfile: linux_ia64_efc+mpi_altix
(I did not find any usefull information in the CVS-check-in messages,
so I am asking the author).
we have currently 3 linux_ia64_efc* optfiles:
linux_ia64_efc
linux_ia64_efc+mpi
linux_ia64_efc+mpi_altix*
But all the 3 have the same comment inside:
#  tested on orion.jpl.nasa.gov (...)
the +mpi_altix is only different from the +mpi by 1 more dir
in the LIBS/INCLUDES/INCLUDEDIRS 
Do you remember why ? (NetCDF ?, MPI ?) ?

Was going to merge at least linux_ia64_efc+mpi into linux_ia64_efc
but might as well incoporate also linux_ia64_efc+mpi_altix into it.

Thanks,
Jean-Michel

On Wed, Sep 07, 2011 at 05:36:56PM -0400, Jean-Michel Campin wrote:
> Hi,
> 
> I would like to clean-up a little bit the list of optfiles (right now, 
> more than 140).
> 
> I have started to merged few optfiles, with the idea of
> having only one optfile for non-MPI and MPI built
> (merging the +mpi optfile into the previously non-MPI one),
> and also doing the same for the OpenMP one (+mth).
> 
> This will reduce the number of optfiles, be easier to maintain,
> and since almost all the specific options are related
> to compiler version rather than MPI/non-MPI choice, it should not
> be a problem. Just need to use genmake2 with "-mpi" and/or "-omp"
> option(s). 
> An other advantage is that, when the optfile is not specified, 
> genmake2 might be able to find the right optfile for an MPI built 
> (at least it was working for me with linux_amd64_gfortran).
> 
> Will do a tag when this merging is done, and then move the old (merged)
> optfiles to build_options/unsupported/
> 
> Also, I think some optfiles which are platform specific could also
> be moved to "unsupported" when the machine is no longer used.
> Will try to make a list and ask if it's OK to remove them.
> 
> Cheers,
> Jean-Michel
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel