More newb questions (sorry). I'm having trouble following the tutorial for running a packaged project. I have made a new directory with a new Kedro project (we expect each client of ours to have their own Kedro project) and have installed the built wheel. However Kedro is looking for nodes and pipelines locally in the new project and can't find the ones in the installed project. Does this mean I have to copy over all of my pipelines manually from the installed Kedro project?

hi @Jon Cohen, I think I understand your setting. running a packaged project is

python -m packaged_project

. running the project from the current working directory is

kedro run

. there are two different commands and they do not mix. what you can do is to

import

the pipelines from

packaged_project

in a way that

pipeline_registry.py

picks them up. so, definitely no need to copy-paste them, but reusability would be achieved by `import`ing packages as with other Python libraries. does that make sense?

I think the issue was that I needed to copy the data and conf folders over and not simply make empty directories like I was reading from the tutorial language. Ultimately I think we are going to be importing micropackages from a GitHub repo, so each client will have a data processing step to get data into a standardized format for us and then they will import a pipeline from GitHub to do the data science.

@Jon Cohen I understand, thanks for the extra context. if you have a moment to open an issue describing the part of the tutorial that didn't match your experience we'd be very happy to have a look. I think this reusability pattern is worth considering.

Sure, I'll write up a quick issue ^_^