Reading parameters from toml in the yaml file assumes that the toml is in the working directory

[Sbozzolo]

Steps to reproduce:

import ClimaAtmos as CA
cd("/tmp")
CA.AtmosConfig(; config_dict = CA.config_from_target_job("sphere_aquaplanet_rhoe_equilmoist_allsky_gw_raw_zonallyasymmetric"))

Returns this errror:

"toml/sphere_aquaplanet_rhoe_equilmoist_allsky_gw_raw_zonallyasymmetric.toml": No such file

As a result, it is not practical to run the various configurations from anywhere but the root of the ClimaAtmos project.

I am not sure what is the cleanest way to fix this. On one side, reading the toml distributed with ClimaAtmos defies the purpose of having them as a configuration. On the other hand, this makes impossible to run the configurations unless one clones the ClimaAtmos repo and runs the code from the root of the project.

[nefrathenrici]

The simplest fix I have come up with: if the TOML path is not a valid file, we can prepend the default toml location to the path, check if that is a file, and notify the user. This would support both:

• users submitting their own parameter files that aren't in the default /toml/ location
• users running the given model configurations outside of pkgdir(CA), who still want to access the default TOML files.

[Sbozzolo]

I think that this would cover most of the common use cases I can think of. (A case that would be not be covered if someone wants to run an example configuration but change one single parameter.)

[nefrathenrici]

A case that would be not be covered if someone wants to run an example configuration but change one single parameter.

Yes, this would require a bigger change, and it is not too difficult in the current setup:

config_dict = CA.config_from_target_job("sphere_aquaplanet_rhoe_equilmoist_allsky_gw_raw_zonallyasymmetric")
config_dict["toml"]  = "path/to/toml"  
# Or config_dict["toml"] = Dict
CA.AtmosConfig(; config_dict)

[simonbyrne]

First step is to document all the input files required.