joshbroomberg / easy-jupyter Goto Github PK

Also move the jupyter_server name to constants

use-local/use-aws probably should not set the home directory. Consider better mechanisms for this.

It is too easy to use-local in one place then switch locations and continue operating on the wrong local directory. There is also a weird state gap if use-local is not called when stop-aws is called but calling use-local during this process seems wrong as it sets the call location as the new home.

Maybe a set-home command which can also be used to blank out the home field during aws teardown.

Switch main script shim and all subscripts to use argbash templates to provide proper shell-script capability and controls. This will facilitate #2 and #4

Add the ability to specify the source of the push and the target of pull. Default to the current folder but consider adding a yes/no prompt to avoid over-writes.

Allow AWS instance size to be set via flags

Given we no longer use env variables for state, it should be easy to run scripts without the bash profile shim. This means we can move to a more standard installation by adding code to the user bin folder.

• On install, copy all files to .easy_jupyter in the home directory.
• Make all appropriate files user-executable
• Add .easy_jupyter to path in the .bash_profile.

Core user flows like AWS should be front and center. Jupyter tweaks etc should be last.

Ensure we have the following:

• Code Wrapping
• No execution order indicator
• Better font and text size

• Persist HOME_PATH to data file to allow for smooth pick up of workflow after terminal refresh.
• Include the use-local command as part of the AWS stop to ensure HOME_PATH is reset properly following AWS shut down.

Ensure themes change the default plotting style when activated.

http://www.gnu.org/software/bash/manual/bash.html#Programmable-Completion

Clean install doesn't write initial state for the variables used in the notebook command producing a broken installation.