| [ease-dev] Jupyter Kernels |
Hi *,
Another quick overview on what happened regarding the Jupyter kernels this week.
I successfully managed to get an echo kernel running. This already is a great step as it means that we are handling almost all parts of the Jupyter protocol correctly.
The code relies heavily on the japyter [1] project by Tobias / OpenAnalytics but needed to be patched to work for the kernel rather than client implementations. I decided to make a temporary fork of the project in order for me not having to worry about breaking their code. During the initial implementation phase I will need to change quite a lot of things, but once the code is stable I will merge it back to their project.
The forked code is already working but I am not happy with a couple of things yet (mostly dependencies). I will change them tomorrow and push them to the github repository so you can see what's happening. Once the code is ready and uploaded it will be easier for me to commit more frequently and in smaller chunks. Up to now it has been quite a mess to get everything running as I needed to change things in several places simultaneously. With the next commit I will hopefully have most parts in a state were I do not need to focus on the underlying messaging protocol but rather implement independent handlers for different types of requests.
I will update you here once the code is online. It already contains a readme on how to get the kernel running within an existing Jupyter installation so you should not have a problem giving it a try.
Best regards,
Martin
[1] https://github.com/OpenAnalytics/japyter