Fix config files

[znicholls]

The config files are currently raw Python files. We should find a better way to do this.

[mathause]

Currently a config state is passed as a param to each function. Also there is no default configuration (or is there?). Can we avoid cfg entirely? If not here are some projects where we could get inspiration from:

• https://docs.oggm.org/en/stable/getting-started.html
• https://docs.dask.org/en/stable/configuration.html

Of course the difficult part is to decide which info should be part of the config and what is passed as arguments.

---

copied this comment from #112 to avoid having a duplicate issue

[mathause]

Some 'config values' I started using.

x_dim="lon"
y_dim="lat"
x_coords=None # -> x_dim (important for RCMs)
y_coords=None # -> y_dim (important for RCMs)
cell_dim="cell"
time_dim="time"

[yquilcaille]

Why having "x_coords" and "x_dim"? Is it called differently in RCMs?

And instead of "cell", why not gridcell? A little longer, but more explicit?

[mathause]

Why having "x_coords" and "x_dim"? Is it called differently in RCMs?

Yes, the coords can be different from the dimensions.

rasm = xr.tutorial.open_dataset("rasm")

x and y are the dims and xc and yc are the coords

And instead of "cell", why not gridcell? A little longer, but more explicit?

No super strong opinion - we leaned towards cell in #105 but I am happy to reconsider.

[yquilcaille]

Indeed, I remember about the differences between coordinates and dimensions :) But I was rather wondering why this mention to RCMs here? Is it different for them? I would have expected that it would have been the same for RCMs and ESMs?

Yes, gp and cell were mentionned, but you were saying it is not a good name. Hence gridcell, but i suppose others may have new ideas?

[mathause]

Yes RCMs are usually on a rotated grid.