Colors in Databricks Notebooks Change Between Dark and Light Mode

[sergiocalde94]

Description:
Colors differ significantly between dark and light modes when using the lets-plot package in Databricks Notebooks. This inconsistency makes it challenging to maintain a unified visualization style.

Steps to Reproduce:

1. Create any plot using lets-plot in a Databricks Notebook.
2. Switch between light and dark mode in Databricks.
3. Observe the changes in plot colors.

Expected Behavior:
Plot colors should remain consistent regardless of the notebook theme.

Actual Behavior:
Colors change depending on whether the notebook is in dark or light mode.

Environment:

• lets-plot version: 4.3.2
• Databricks Notebooks (DBR 15.1)

[alshan]

That is surprising. As far as I know, Lets-Plot is not aware of Databricks color profiles. Most likely, some kind of glue code was added by Databricks or a third party. I'm not familiar with Databricks notebooks, but perhaps there is an option to disable this behavior.

In lets-plot colors are adjusted either via theme() or by applying a "flawor", for example: flavor_darcula.

Just as an experiment, try to set the global theme in your notebook:

LetsPlot.set_theme(flavor_darcula())

and see if this would fixate the plot colors.

Unfortunately, we don't have a "flavor" that reverts the plot colors to default values. However, we can add this option in the future if this is your intent and the trick above works.