common-lisp-jupyter
common-lisp-jupyter copied to clipboard
yitzchak
Possible 'emacs-ipython-notebook' interaction problem?
In the notebook in browser, I get results from running a cell. However in emacs, with the 'emacs-ipython-notebook' the websocket here is reporting an error of (json-end-of-file) after running the cell. Uncertain if this is due to jupyter / this kernel / ein or some mix there in.
It doesn't seem to be ein, as it handles IHaskell fine. Is there a place I should look in this kernel to see why it is obviously giving the output to jupyter in the browser but a websocket from emacs is not getting the same info?
Unfortunately I have absolutely no experience with emacs let alone with emacs-ipython-notebook. It sounds like a JSON encoding error on either the EIN or the common-lisp-jupyter side.
The argument the IHaskell works fine therefore EIN must not be the source of the error isn't really conclusive since common-lisp-jupyter works in Jupyter Notebooks, Jupyter Lab and it passes the kernel tests published by the Jupyter people.
I suspect that I'll have to setup EMACS and try it out to trace down the issue, until then can you give me a minimal example? Does it fail evaluating nil?
Minimal example would be nil, one attempt at executing a cell with only nil fails with:
ein: [info] WS action [(json-end-of-file)] on-message (ws://127.0.0.1:8888/api/kernels/dfab0125-b53b-43cd-baff-c491e54660b2/channels?session_id=7c13390a-d129-4454-abe2-1f9f436d08ae) [5 times]
It might be that his JSON parser is being more strict or something. I will look into that as well. I did just notice the [5 times] bit, being uncertain about how these things communicate, does this indicate that jupyter sent 5 objects? Or was is this possible an ein internal thing, in that since it had an error ein asked the service to send it 4 more times in hopes of a better reply?
