Quick Q, we can create our custom starters, which is very useful - we use it to set company-wide standards. Is there a way to create custom pipeline layouts? So when we do

kedro pipeline create

it will use a custom cookiecutter layout created by us? Technically it seems possible as I see in the

kedro

repo there is a dedicated cookiecutter for pipelines

Very interesting question!

I'd say the concept is very similar to the sub-folders of the

data

directory. Very much the same idea of semantically ordering data, we also have a semantic ordering of nodes, so I want to reflect that in our organization's starters

So you want kedro pipeline create not just populate an empty pipeline file, but with pipeline and nodes - then the user should update and fill in the definition of these nodes?

Specifically instead of having all nodes in

nodes.py

I'd have them split, either

nodes/xxx.py

or just having multiple

xxx.py

files

I will check later - I don't remember if this is possible or not.

You can use cookiecutte directly for this if you want to get going quickly

1. How would I integrate the cookiecutter with the

kedro new pipeline

command? 2. Will the pipeline discovery and other features related to pipelines will just work?

1. It wouldn't be via kedro, you'd use it natively

2 depends on the pipeline.py if i remember - the docs have more information about what structure is expected. It check the existence of pipeline.py and the create_xxxxxxxxx method.