I'm confused by the kedro kubeflow integration... 1. The main kedro project talks about using a

build_kubeflow_pipeline.py

python file to compile the pipeline to a yaml file that is then uploaded. But the kedro-kubeflow plugin has a

kedro kubeflow compile

function that uses

conf/base/kubeflow.yaml

to generate the pipeline yaml. Is compiling from

kubeflow.yaml

the correct way to go here? 2. I'm trying to get my pipeline to compile using

kubeflow.yaml

but

kedro kubeflow compile

seems to want to upload to the

host:

too... which seems odd for a compile semantic. In my case, I don't have cluster access except through the UI... so how can I truly just have it only compile?