| [ease-dev] Jupyter Kernels |
Hi,As I already wrote, last week I was working on getting the execution part of the Jupyter kernels to run. Everything looks good so far, there are just a couple of things I might need to optimize but the execution itself is running pretty smoothly.
Today I worked on publishing data from the kernel back to the clients. I introduced the new Interface IJupyterPublishable that is used to return data back to the client. As mentioned it is possible to send different representations of the same object. I implemented a quick dummy function that returns a png image both as the byte data and a text representation. As everything is mapped using MIME types, the notebook picked up the data immediately and displayed the picture. As the console client only has text based UI, it ignored the image data and used the text representation instead. This is exactly the behaviour we want and I am glad that everything could be integrated smoothly. I am still working on the adapter framework to allow additional adapters to cast known objects into their corresponding representation format. I hope to finish this in the next couple of days. I will keep you posted here.
Once the adapter part is finished as well, I will go over the code once more and maybe refactor the parts that are not implemented fully optimal. Overall I am pleased with my progress so far and the current results look very promising.
Best regards, MartinPS: I hope it is okay that I will publish my progress on Mondays from now on, as I usually use the whole day for GSOC.