Jupyterlab BokehJS loading wonkyness and CSS Theming

I noticed something annoying via the PyViz/HoloViz tutorial this week at Scipy and filed Panel labels do not follow jupyterlab theme · Issue #520 · holoviz/panel · GitHub

I started looking to see if the jupyterlab_bokeh extension does anything to hook Bokeh’s CSS into jupyterlab’s theming, mainly to rule out the issue I found going away if I also installed the jupyterlab_bokeh extension and I noticed that jupyterlab doesn’t seem to be updating the notebook output area with “/[Logo] BokehJS x.y.z successfully loaded” although it does seem to be rendering the 2 non-embedded-server notebooks in the test_cases dir of bokeh/jupyterlab_bokeh repo.

  1. Do you think this failure to update the jupyterlab cell output is another symptom of bokeh/jupyterlab_bokeh#49, if so, I’ll update that ticket. If not, is this worth having another ticket?
  2. What do you think about having the jupyterlab_bokeh extension provide CSS that hooks Bokeh CSS upto the jupyterlab CSS variables that can be set by jupyterlab themes? I’d be happy to work with the Jupyter people here at SciPy to craft a PR for you.
  3. It’s a bummer you didn’t make it to SciPy this year. Hope you’re doing well.

Hi @Tim_Snyder First off as of #8989 Modularize CSS there aren’t actually separate CSS files anymore. (We still publish 0-length files to CDN to keep the current resource loading code identical, but those will go away with Bokeh 2.0) I am not sure how or if this intersects with you observations. I also don’t know offhand if this is related to bokeh/jupyterlab_bokeh#49. Since there is not alot of issue traffic, I would suggest for now creating a new issue that describes everything but also linking the two issues with comments.

Thanks for looking in to this, I am not really a JLab user and I am fairly at max capacity for commitments, so any and all help is appreciated. FYI if anyone wants to sprint on Bokeh, I can make sure to be around some, e.g. on the Gitter channel.

Hopefully can make next year!

Thanks for pointing me back to #8989. I noticed it but didn’t read it carefully. I see where the CSS is coming from now.

I’ll see what I can figure out.