Calling BokehModel in vscode (interactive cell) results in Error: No version of module @bokeh/jupyter_bokeh is registered

[samimia-swks]

python 3.12 on Ubuntu (vscode + WSL)
bokeh v3.6.1
jupyter-bokeh v4.0.5
jupyter-core v5.7.2

I am trying to push updates to a bokeh figure inside a vscode interactive enviroment (.py file with cell tags) using these instructions:
https://stackoverflow.com/questions/77841047/how-to-update-a-bokeh-plot-inside-vscode-jupyter-notebook

So basically :

from bokeh.plotting import figure, output_notebook
from IPython.display import display
from jupyter_bokeh.widgets import BokehModel

output_notebook()

plot = figure(title='Update Example', x_axis_label='Time', y_axis_label='Value')
line = plot.line([], [])
handle = display(BokehModel(plot))

This results in the following error

No version of module @bokeh/jupyter_bokeh is registered

and a vscode pop up with the message:

Unable to find widget '@bokeh/jupyter_bokeh' version '^4.0.5' from configured widget sources ["jsdelivr.com","unpkg.com"]. Expected behavior may be affected. Click here for more information.

[MarcSkovMadsen]

I have simulator issue and not Ben avle to use panel in VS code for a while.

[jwintel]

@samimia-swks @MarcSkovMadsen I had a similar issue and found the answer here: the link at the bottom of the top post "Click here for more information" contained settings that weren't in my vscode.

Once I added "jupyter.widgetScriptSources": ["jsdelivr.com", "unpkg.com"], to my vscode settings widgets started rendering in the interactive vscode for me.

It looks like the original poster already had these settings, so I'm not sure why it's not working in their case.

[FalseFeint]

I'm running into this as well. In my case I'm running a jupyter notebook on a HPC and when I'm on a login node with internet I'm able to render my plots fine. However, when I run my notebook on a compute node that does not have internet, I see the same behavior that @samimia-swks shared as well as the pop up saying @bokeh/jupyter_bokeh can't be found from jsdelivr.com or unpkg.com. Because I don't have internet, I won't be able to connect to either of these CDNs.

Is there a way to install/register @bokeh/bokeh_jupyter so that I don't need to access these CDNs in the first place?