Ivan Ruiz Manuel

**Edit: see updated setup below.** After some discussion between @brynpickering and I, this is the setup I've come up with. It's meant to enable easier debugging. The most important feature...

Edit: I no longer recommend this approach to `config`. See below. ~~@sjpfenninger I mostly agree. ~~ - ~~I somehow feel like `_config_from_init_overrides` may be too much, and perhaps we should...

After working on the first parts of this (#639, #641), I think that separating `model.config` into `config_file / kwargs / etc` is a TERRIBLE idea. It introduces many anti patterns...

Here is the newest version of this proposal. | Old setup | New setup | Comment | Type | |-------------------------------------------------------------------------------------------------|-----------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------|--------------| | `.math` `._model_data.attrs["math"]` `._model_data.attrs["applied_additional_math"]` | `.backend.math` | New handler class...

@brynpickering: 1. If we want users to "see" something, we have to make pace with the fact that they can mess with it. It's the Python™ way - For internal...

@brynpickering I am thinking that perhaps this is something that our `read_rich_yaml` functionality should take care of... This way the `pydantic` model always expects lists in every location, and we...

@brynpickering honestly, after going through this whole schema thing and the math bits, I am not sure how beneficial this would be. This could also add spam in cases where...

@brynpickering Correct, that should work well! This is a chicken and egg problem: either the templates define the yaml/tables, or the yaml/tables defines the templates. Resolving the YAML logic first...

@brynpickering I believe @jnnr is only referring to `active` now, though! Defining `active` should be possible in the tables, I think... It's just that you'll need different files if you...

I did not really meant it as striving to be pythonic (`a = b` is assignation in all programming languages I know tbh), but more about the consistency and readability...