This message was deleted.

I don’t think you need two .yaml files. As you describe it, such a pipeline is already supported. Perhaps share your .yaml file or describe your pipeline in more detail, and explain where you think ploomber falls short here?

Just to clarify, you want something similar to this: https://docs.ploomber.io/en/latest/api/cli.html#build-pipeline-partially Where it’s pointing to a different end?

great question, my personal preference is that conditionals inside a pipeline get a bit hard to maintain. however, there are a few alternatives I'd suggest: 1. partial builds can allow you to select the final node to execute, so you could so

ploomber build -p random-forest

or

ploomber build -p another

2. alternatively, you may parametrize a task and have that parameter determine the logic to execute inside the task. example:

tasks:
  source: script.py
  products: ...
  params:
    model: random-forest

then in your script.py:

if model == 'random-forest':
  # do something
else:
  # do another thing

3. Finally, if the logic is too different, you can create the common parts in a separate file and then use the

import_tasks_from

directive https://docs.ploomber.io/en/latest/api/spec.html#import-tasks-from thoughts @Raffaele Olmeda?

Yes that was exactly what I meant, thank you guys!

happy to help! meerkat

what if in his piacture, I have a task F which is common to both branches (after D and E)

you could create two tasks that have

source: F.py

, just add a

name: some-name

to each one

Annnn ok so it works also in this case, thats great!

Feel free to ask more it helps us improve the product and docs 🙂