Open MarcSkovMadsen opened 2 years ago
Can pure Bokeh plots be rendered in code-server?
Yes. There seems to be no problem in the interactive window. But the notebook is problematic.
For me the interactivity in Code is key. It is hard for others to keep up with and compete with that environment. So need to work there.
This is an example of Bokeh/ Holoviews plots being render in code-server @maximlt . I'm used to having full interactive with 2-way comms in VS Code on my laptop. Here it does not work. But I can get it to render in the Interactive Interpreter. In the Notebook everything gets too slow.
I'm trying to create a nice repository for me and Sophia M Yang for a blog post and tutorial. I would also like to show case that the HoloViz ecosystem works with VS Code. So I added the code-server to the Binder environment.
Unfortunately it seems not to work
Reproducible Steps
https://user-images.githubusercontent.com/42288570/149618402-0d65e26d-c85d-435c-8c28-8a6b34981754.mp4 (The video is speeded up with a factor 1.7)
Repo: https://github.com/MarcSkovMadsen/hvplot_interactive commit 3cd8bf19dc326a76efe20e2dd3fedb7f1ff6e6af
notebook
Notice how it took a long time to execute this cell
Notice the console how you get an error message as you print hello.
In the video this load pretty quickly (8secs) but I've also experienced it running for minutes before I've given up.
It stopped working.
Versions
jupyter_bokeh bokeh Version: 0.12.6 Name: holoviews Version: 1.14.7 Name: hvplot Version: 0.7.3 Name: ipykernel Version: 6.7.0 Name: jupyter-bokeh Version: 3.0.4 Name: bokeh Version: 2.4.2 python Python 3.7.12 Python VS Code extension Version: ms-python v2021.12.1559732655 Jupyter VS Code extension Version: ms-toolsai v2021.11.1001552333 code-server Version: 4.0.1 735c6da829535969ff7193c79379299e4a1cb9bc
Log for Output - Jupyter window
stacktrace.txt
Additional Context