We are running JupyterLab v3.6.7. In my lab the autocompletion is breaking. It may work once or twice, then stops. When broken I can see the Kernel Indicator blink, so something is trying to happen. Other labs on the same server are not having the same issue.
Is there any way I can view/enable logging from my lab? I only have access to my lab and not the server itself.
This might be an issue with ipykernel. I would check if you are running the latest version and if not try upgrading (and if yes, consider trying to downgrade). I would also check if disabling jedi by running %config IPCompleter.use_jedi = False. Please do let me know if either of these helped (and which version you were using).
The logs for ipykernel failures will unfortunately require access to server logs.
We are running Python 3.10.6. IPython is showing v8.5.0. I’m not sure that those are the version numbers you are looking for.
Unfortunately, even if I run an empty cell, the autocomplete breaks. If I try to run your command, autocomplete stops working.
When I start my lab, autocomplete works. If I do anything in my lab, the autocomplete breaks and won’t work until I restart my lab. By “anything” I mean, run an empty cell, add a new cell, or try to autocomplete on a non-existent variable.
When I try to autocomplete, the round kernel indicator in the top right corner does flash, so something is trying to happen.
I’m going to talk to admins to try and get access to the logs… I’m sure that there is an error message is there somewhere that would be helpful. Is there any file in particular I should be asking for?
I just found something interesting… Under the Help menu there is the “Launch Classic Notebook” option. If I use that, the autocomplete does not break.
Under Help->About it shows Jupyter Server is v1.21.0.
One more item to note…
When my autocomplete is broken, I can still use Shift-tab to view the documentation on a command.