Luca Bertagna

Infrastructure clean-up and simplification tasks

1

As we approach a complete v1, we are realizing that some parts of the infrastructure are clunky and/or overly complicated (like adding lots of code/logic to support scenarios unlikely to...

Add possibility to avoid calculations in atm procs for certain diag fields, if they are not outputed

1

There are some fields computed by some parametrizations that are diagnostic, and not needed by other atm procs. We could devise a way to avoid allocating/computing these. E.g., adding a...

Should we maintain SCREAM_HACK_XML to allow manual edits to namelist_defaults.xml

3

Currently, if you do `./xmlchange SCREAM_HACK_XML=TRUE`, you can then manually edit the namelist_defaults.xml file, and be sure that buildnml won't overwrite it (in particular, any atmchange call will have no...

Add baseline testing for standalone EAMxx

2

Our struggle tracking down the problem in #2534 suggests we need to improve our ad-driven tests. In particular, we do not check for consistency across runs. For CPU builds, we...

I noticed that there are a lot of pack sizes configurable options, though most are completely unused. It may be worth it to clean up a bit, for clarity/maintainability. In...

I am opening this issue to track stuff that still needs to be done for our handling of namelist in v1. Feel free to add items that you think are...

Make atmquery print all XML children if node name is not a leaf

1

Given a node like ``` shoc enable_column_conservation_checks: false number_of_subcycles: 1 enable_precondition_checks: true enable_postcondition_checks: true repair_log_level: trace internal_diagnostics_level: 0 compute_tendencies: NONE ``` we currently get ``` $ ./atmquery p3 namelist_defaults::atmosphere_processes::physics::mac_aero_mic::p3: $...

There are some fields computed by rad that are only relevant in certain areas (e.g., on cols where it's day), and that are set to something like "fillValue" elsewhere. When...

This PR removes CMake warnings like the following, which are polluting the output quite a bit. ``` CMake Warning (dev) at /home/lbertag/workdir/utils/cmake/cmake-install/share/cmake-3.27/Modules/FindPackageHandleStandardArgs.cmake:438 (message): The package name passed to `find_package_handle_standard_args` (PnetCDF_C)...

FindNetCDF.cmake should not ignore `NetCDF_<COMP>_PATHS` if `MPI_<COMP>_FOUND` is TRUE.

9

The helper function `initialize_paths` completely overwrites the user-provided `NetCDF__PATHS` variable if `MPI__FOUND` is TRUE. Instead, it should _append_ additional paths to it.