Aart Goossens

@maresb This works! You are my hero! :superhero: Awesome. I think I tried something similar by passing `sqlalchemy.orm.relationship()` directly to `sqlmodel.Relationship()` as the `sa_relationship` kwarg but I must have missed...

That's a good point. I'll reopen the issue.

@Iionsroar I think this is not related to this issue, but this one https://github.com/tiangolo/sqlmodel/pull/322.

A. Good idea to make working with strava data as easy as possible. I'm not sure if I would want to add this to the wdf class because I like...

A. I don't think I agree with your point. An end user that wants to import data from strava would just use sweat.io.strava and get a wdf in return, he...

With your analogy my approach is identical to pandas: IO methods are not on the dataframe class but are separate methods in the pandas/sweat library: `pandas.read_csv` vs `sweat.io.strava.get_activity`. As far...

1. I think end users shouldn't have to deal with importing from `sweat.io.models`, for example `sweat.io.goldencheetah` should handle that. The end users just imports `sweat.io.goldencheetah.get_latest_activity()`, calls it and gets a...

Good idea, in your proposal both work flows are supported without adding much nesting to the modules. So then the structure for `sweat.io` would be?: ``` +-- io | +--...

Good point about code in `__init__.py`: It's good practice not to have code there. Importing from __init__() could work or if we want to be really strict about we could...

Thanks @piotr-kubica. Yes I agree that enforcing