Handle long-running computation use case

[vertexcite]

If a computation will run for a long time, it would be good to be able to disconnect the browser, and reconnect later, and see any output produced during that time (Similar issue for Jupyter see jupyter/notebook#1647 ).

This is especially applicable if you are running the server on a separate computer (e.g. cloud host with GPU), and you don't want to have to keep the client computer (e.g. your laptop) on and connected the whole time the server is computing.

[NickSeagull]

Great, thanks for the idea!

[harveyslash]

Somebody mentioned using a proxy browser to communicate to jupyter (those will keep a constant connection) and then using a local machine to see the contents. Are there any tools/projects that can allow this? I am unable to find anything that would allow me to do that.