Atmosphere/physics makefile updates #1238
Conversation
…tatements "use mpas_derived_types" and "use mpas_pool_routines", removed "use mpas_derived_types" which is not needed. * In ./src/core_atmosphere/Makefile: -> Cleaned up all the dependencies. -> Moved compiling mpas_atmphys_o3climatology.F from OBJS_init to OBJS. -> Removed the include -I../../operators which is not needed.
-> moved mpas_atmphys_interface.o from OBJS_dyn to OBJS. I recall that originally,
we used OBJS_dyn to handle separate interfaces between the dynamics and physics
when we had the hydrostatic and non-hydrostatic dynamical cores. Now that we
only run the non-hydrostatic dynamicol core, we can simplify Makefile.
-> consequently, also removed core_dyn.
|
@ldfowler58 I think I stumbled across a couple of other missing dependencies in I can add a commit with these changes to your branch, or if you'd prefer you can fold these changes into your existing commits. |
|
I also have some changes that are mostly confined to the physics that should enable us to run parallel builds of MPAS-Atmosphere with, e.g., |
…e dependencies for module_ra_rrtmg_sw.o.
-> In module_mp_thompson.F, specify that the functions gammp,wgamma,rslf,and
rsif are the only functions to used from mpas_atmphys_functions.
-> In module_mp_thompson_cdfra3.F, now use mpas_atmphys_functions instead of
use module_mp_thompson since we only want to use the functions rslf and
rsif from mpas_atmphys_functions.
|
I corrected the dependency of module_ra_rrtmg_sw.o in the physics_wrf Makefile. Thanks for pointing this out to me. About the dependency for module_mp_thompson_cldfra3.F, the module should really use the functions rslf and rsif in mpas_atmphys_functions.F and not the functions rslf and rsif from module_mp_thompson.F. Note that in module_mp_thompson.F, those two functions are bracketed with a #defined (mpas) statement so that they are removed by cpp and would not show in module_mp_thompson.f90 with GEN_F90=true. In module_mp_thompson.F, in "use mpas_atmphys_functions", I now request to only use the functions gammp, wgamma, rslf, and rsif. I am interested in looking at the changes that are needed in physics to allow a parallel build. It may be better to add them in a separate PR? |
.gitignore
Outdated
| @@ -3,10 +3,12 @@ | |||
| # All pre-processed Fortran files in MPAS specific directories | |||
| src/core_*/Registry_processed.xml | |||
| src/core_*/*.f90 | |||
| src/core_*/*/*.f90 | |||
There was a problem hiding this comment.
I think it would be cleanest to just add a rule for *.f90 in .gitignore files in the src/core_atmosphere/diagnostics and src/core_atmsphere/utils directories. As a separate framework issue, I'm actually wondering whether the top-level .gitignore file should be dictating which files are ignored within core subdirectories at all.
|
OK. I will remove the last commit regarding .gitignore and push the updated sourcecode to github. |
ldfowler58
force-pushed
the
atmosphere/physics_Makefile_updates
branch
from
1f5a38e to
8737b09
Compare
Clean-up of Makefile in physics directory
In this PR, we removed the statement "use mpas_derived types" in all the mpas_atmphys modules where we also had the statement "use mpas_pool_routimes". We then proceed to clean-up and simplify Makefile.