Tim Head

I'm looking at [Empirical evaluation of the impact of k-means initialization](https://scikit-learn.org/dev/auto_examples/cluster/plot_kmeans_stability_low_dim_dense.html#empirical-evaluation-of-the-impact-of-k-means-initialization)

Hi 👋 nice work on picking up this issue and making a PR. Three things to think about adding: 1. a test the checks that the exception is raised when...

Thanks for updating the PR. You are right, because `indices` is a numpy array we can't use `if not indices`. Looking around the code base I couldn't find what the...

Maybe the existing plugin mechanism (https://papermill.readthedocs.io/en/latest/extending-entry-points.html#developing-new-i-o-handlers) is something to try out in order to support different input formats (right now they are mostly used to read from different kinds of...

> One more question: should the plugin be a separate Python package, or do you expect it to become part of Jupytext itself? No idea. I think having it be...

Why not use the `TextFileHandler` class that was suggested in the previous comment for loading and saving Jupyter notebooks stored in a different format?

Some things I think don't need to be solved by papermill are situations like: ```python def foo(a, **kwargs): pass foo(1, a=3) ``` If you end up in a place where...

> Though passing args down the stream as a dict might bring other issue when we'll have to merge 2 or more dicts - so back to the original problem....

I think notebook authors can already use mlflow to track their parameters and models when they are executed with papermill. Do you think the `papermill` Python library should also provide...

👍 for docs on how to do this in JupyterLab. For classic jupyter notebook: