Based on earlier conversation with @fperez and @ian-r-rose
When a notebook is being executed automatically (non-interactively), it would be really cool to be able to open up the notebook and watch it execute in real time, and perhaps play with visualizations as they're generated. If an exception is raised, rather than the kernel dying, control can be passed to the user, who can resume automatic execution after some debugging.
As in JupyterLab Google Drive, except actually logging a machine in to a google account is probably not Kosher. This would take some thought, and would likely reuse large pieces of the drive plugin, but with a custom API to replace the Google API.
Based on earlier conversation with @fperez and @ian-r-rose When a notebook is being executed automatically (non-interactively), it would be really cool to be able to open up the notebook and watch it execute in real time, and perhaps play with visualizations as they're generated. If an exception is raised, rather than the kernel dying, control can be passed to the user, who can resume automatic execution after some debugging.
As in JupyterLab Google Drive, except actually logging a machine in to a google account is probably not Kosher. This would take some thought, and would likely reuse large pieces of the drive plugin, but with a custom API to replace the Google API.
This is definitely a long-term goal.