Having gadi_jupyter as a bash script is becoming unweildy. Consider splitting into two halves - a frontend written in bash (for portability) that users run on their own computer, and a backend written in python that runs on gadi that does the actual queue submission.
This would allow us to handle things like resources for different queues in a more sensible language
Having gadi_jupyter as a bash script is becoming unweildy. Consider splitting into two halves - a frontend written in bash (for portability) that users run on their own computer, and a backend written in python that runs on gadi that does the actual queue submission.
This would allow us to handle things like resources for different queues in a more sensible language