the default behavior should be to load from a repository config file which documents where to look for the data

[georgemilosh]

Climate-Learning/ERA/ERA_Fields_New.py

Line 1722 in 83f1578

self.datas = xr.open_dataset(ut.first_valid_path(self.mylocal,self.filename))

[AlessandroLovo]

Can you please elaborate on this?

[georgemilosh]

Basically as I started working from other clusters, the default path doesn't seem to be reasonable. Thus there is no reason to hard-code it inside the Learn2_new.py, that is to say, its default value doesn't make sense outside of CBP or if anyone else outside would use our software. So I would recommend having a separate json file that just contains the default folder. Default name could be the name of that json file for example.

[AlessandroLovo]

Ok this makes sense

[AlessandroLovo]

Now that we want to work with Reanalysis data as well this seems even more important

[georgemilosh]

So you are not happy with my CESM hack? See CESM_tuto.ipynb

[AlessandroLovo]

On the contrary, I do like it. But it is indeed a hack. At some point we should make things more natural maybe.
With Valeria we wanted to use the Reanalysis data, but right now all the interface (for Plasim and CESM) is hardcoded in Learn2_new and ERA_Fields_New.
So, maybe, this is a good time to make things nicer.
However, it's faster to just make another hack like yours, so we'll likely go for this approach for the moment 🥲

[georgemilosh]

Well you have evaluate how much time you both have and prioritise what is important for you. Having pretty repo is nice, but it should also be not overly complex to be comprehended in case something needs to be changed.

[AlessandroLovo]

Indeed