Hamilton Open Source

image.png

~Running the code locally to reproduce — on first glance I agree that something is off…~

```def full_df(add_model_col: pd.DataFrame) -&gt; pd.DataFrame:
    pass```
should be:
```def full_df(add_model_col: pd.DataFrame) -&gt; pd.DataFrame:
    return add_model_col```

with `subdag` we now need the body to return what was input (when it was `reuse_functions` we didn’t).

I still get the same results even when I return `add_model_col` as above

Awesome! Something to point out — the two DAGs above are a byproduct of how its implemented. It displays *both* the final DAG as well as the extra subdag. If you put the subdag in a separate module/don’t include it in the driver, there will be only one. It doesn’t make a huge difference as you only get what you execute (and names are unique), but just in case you were wondering :slightly_smiling_face: