Kedro is an open-sourced Python framework for creating maintainable and modular data science code.

Kedro

Hi :slightly_smiling_face:

I am running a simple pipeline which has the following config in a yml

```simple_conn_pt_model_filter_predict:
    date_column: date
    window_length: 0d
    gap: 0d
    check_groups: null
    continue_if_missing: true```
I am trying to edit the parameter `gap`  through `kedro run --pipeline ... --params=...`, but I need to overwrite the whole dictionary

This is because the dictionary is passed as a whole to a node, and updating that specific subkey will not actually update the whole dictionary

Is this a reasonable way to operate? If we had more robust deployment patterns I could simply replace the specific config.yml file, but atm it isn't trivial to do

Using the `--params` flag as you've shown above is a reasonable way to operate, especially when you only have a few parameters to update. Just be aware that you need to provide all the key-value pairs for the new dictionary, even if only one key-value pair is changed.

I was having issues getting kedro to recognize the parameters as a whole dictionary, I think it was parsing it as a string. I ended up going for modifying my deployment a bit and injecting an additional parameters file in my kubernetes pod through the configMap :slightly_smiling_face: