Commit CRTMv3 configuration (backward compatible with 2.4)

[karpob]

Description

Adds CRTMv3 capability while maintaining 2.4 compatibility. Also modernizes setup removing deprecated (or soon to be deprecated) "python setup.py install" method in favor of "pip install . " method. Adds helper scripts to install CRTMv3 + pyCRTM, along with dependencies using miniconda.

Also resolves #17 fixing issues related to setuptools version changes, along with f2py issues. This is done by replacing f2py executable which has bug related to passing preprocessor directives with "python -m numpy.f2py" which passes preprocessor directives correctly.

Issue(s) addressed

Resolves #17

Dependencies

List the other PRs that this PR is dependent on:

• none

Impact

Expected impact on downstream repositories:
None

Checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation
• I have run the unit tests before creating the PR

[BenjaminTJohnson]

@karpob as an FYI: future CRTMv3 after v3.2.x will be netCDF only.

[karpob]

Yeah my thought is when it's netcdf only to move all of that out of setup.py into something like cartopy does for higher resolution maps where it downloads on request when it doesn't find anything. Also, at some point figure out how to remove setup.py completely.

[chengdang]

Hello Bryan, This is cool! I like make_it_so.sh very much. I used ./make_it_so.sh apple_silicon with M2 chip

There are a few hiccups though that I had to manually install some libraries, which seems to be caused by the failure of activating a virtual environment with the following command.

export CONDA_VENV='pycrtm'
conda create --name ${CONDA_VENV} python scikit-build h5py netcdf4 gfortran libnetcdf netcdf-fortran meson cmake git git-lfs matplotlib
Channels:w
 - defaults
Platform: osx-arm64
Collecting package metadata (repodata.json): done
Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - netcdf-fortran

Current channels:

  - defaults
  - https://repo.anaconda.com/pkgs/main
  - https://repo.anaconda.com/pkgs/r

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

I'm able to install netcdf-fortran with conda forge though:

conda install conda-forge::netcdf-fortran

All other libs work with conda create.

[karpob]

@chengdang Good catch! I forgot to move my .condarc when I was testing this (I usually just put conda-forge by default by habit to avoid licensing issues.) I added a line conda config --add channels conda-forge that seems to take care of it when I ran a test (duplicating your issue, then fixing it with the extra line).

[chengdang]

Thanks for the quick fix! Everything works as expected now. I am able to create pycrtm, and run the example. Thank you!

[karpob]

I added back the --fcompiler switch as it's necessary for older builds (spack 1.6.0, trying to build with intel, for example). Hopefully we don't get to a point where they update f2py to throw an error instead of a warning.

[karpob]

Also added fix per #20 where LD_LIBRARY_PATH wasn't set automatically as designed, when using shared libraries.