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

Kedro

image.png

I'd say yes! (But I very much fear this is a trap :face_with_monocle:)

It’s actually a valid one:grin: (which I thought is not). I have always have the impression that it need to be a DAG, turns out it can be more than 1.

This actually makes my life easier as I am working on a solution to recover a failing pipeline.

This makes me wonder, is this mapping allowed in the orchestrator. i.e. Can you run this on Airflow or does `kedro-airflow` work in this case?

kedro-airflow will run, but I think Airflow does not accept tasks without dependency, is it similar in Argo? <@U03S12LHNNQ>

I believe this is a valid DAG though? There’s no requirement for a DAG to be fully connected - see <https://stackoverflow.com/questions/19758373/need-some-clarification-about-dag-directed-acyclic-graph>.

<@U03RJ2PH79D> Sorry, even after looking it up (and without trying it), I'm not 100% sure? You can have multi-rooted dags, but I'm not 100% sure if it can be disconnected. Even if it's not possible, could have a second `dag` type? I think it's mostly just for organization, as DAG is only a subset of things you can schedule with Argo Workflows.