I’ve been playing with the Remote Execution smart cell, and it’s been mostly fantastic. Just about everything I was looking for when I made this offhand comment back in May.

I’m left with two major questions:

1. I’m working on creating a LiveBook that will be mostly focused on interacting with one specific application. I’d like to be able to set the node name and cookie somewhere (maybe in the setup; maybe in a regular cell) and have those become the values (or at least the defaults) used for each Remote Execution smart cell I add. I’d be happy to use secrets for them[1], but it looks like you can’t use a secret or variable for the node name. I suppose the alternative would be to create my own smart cell that uses my target runtime’s settings instead of allowing for user input.

2. I’ve found that I get an error if the Elixir/OTP versions of the livebook environment and the remote are not exactly the same, down to the elixir point release (i.e., Livebook built on 1.15.2-otp-26 while my remote runtime is 1.15.4-otp-26). Is there a way around that, or is it just how it is? I don’t mind building a livebook container using the same base image as my app, but if there’s another way, I’d love to know it.

[1] Side question: Is it possible to programatically set a secret from a cell (or in the notebook setup) for use in another cell?