Matthew Seal

> there's places in the codebase where MultiKernelManagers are referred to as KernelManagers, making the confusion more, so be careful with that distinction. I would really like to correct these...

Potentially yes. But I do worry that proposal has not worked through ~a lot~ all of implementation details or complexities it would introduce as it currently stands, so it's not...

I think this is a good idea, might need implementation back in nbclient to support it. I'll look a bit deeper into what's needed.

So most nbclient configurations are passable from papermill's execute_notebook interface but the kernel_manager's config is not exposed on the nbclient `__init__` interface so those nested config attributes are not modifiable...

If there were done then we could send that along to nbclient: ``` papermill.execute_notebook('Untitled.ipynb', 'Untitled-out.ipynb', config_file='./path/to/my/config.py); ``` or from the cli something like: ``` papermill Untitled.ipynb Untitled-out.ipynb --config_file='./path/to/my/config.py' ```

As to supporting kernel messages to delay shutdown, that could work but it'd probably be more at the jupyter protocol contract level to support a new message pattern to reset...

That would also work. I started with posting this originally (which does the same thing without a new engine, but doesn't work from the CLI): ```python class LongerWaitAsyncKernelManager(jupyter_client.AsyncKernelManager): shutdown_wait_time =...

Registering a cleanup cell with a magic sounds useful

Hmm getting stuck for 20 minutes sounds odd. I've not encountered such a slowdown that was caused by papermill. You mind posting a notebook (maybe trimmed down?) that reproduces the...

I agree with @betatim that the plugin system was made to support this. If you get that plugin working and it has good general value well I'd be open to...