Do we want to allow command-line overriding of parameters from `parameters.in`?

[axch]

Then the semantics of parameters.in becomes that it's the default configuration, and command-line arguments take precedence.

Considerations:

• Probably mostly relevant during fast-iteration exploration, or when automating multi-run batches.
• Pro: Becomes easier to vary model runs, exploring subsets of parameter space.
• Pro: Becomes easier to automate varying parameter space.
• Con: This would be pretty annoying to implement in the Fortran core, but not too bad in a Python wrapper.

For the sake of reproducibility, would want the program to dump its full configuration to a human- and machine-readable file.

This becomes more useful in conjunction with automatically computing the gridded input files from resolution-dependent sources, so that the model resolution can be easily varied.

[edoddridge]

I can see this could be useful, but how much value do you think it adds beyond us providing an example python script that automatically edits and creates a bunch of parameters.in files in appropriate directories?

[axch]

If I do this at all, I was thinking of making it be a feature of the Python wrapper program rather than the Fortran core. In that case, much of the content of example vs implementation will be the same.

[axch]

Also agreed that this is not very high priority.