An experiment in bridging the gap between Jupyter Notebooks using ipywidgets and a Bokeh server application. The idea is to test if this might be a good way of sharing prototype apps built in the notebook.

In this experiment the jubo module implements methods/classes from ipywidgets. If used in a notebook context these will simply be the ipywidgets objects, in a Bokeh context they will be Bokeh objects wrapped to implement (a very limited set of) the ipywidgets interface. Additionally there is some patching of IPython to capture output to be displayed and render it in Bokeh.

• jupyter-lab
• ipywidgets
• bokeh

A verbose requirements.txt file is also included in the repo.

Enable the ipywidgets Lab extension:

jupyter labextension install @jupyter-widgets/jupyterlab-manager

Start Jupyter Lab with jubo on the PYTHONPATH:

PYTHONPATH=../:$PYHTONPATH jupyter lab --notebook-dir=notebooks

View and run simple-interactive.ipynb or images.ipynb

convert to python

jubo/convert.py notebooks/images.ipynb bokeh_apps/images.py or jubo/convert.py notebooks/simple-interactive.ipynb bokeh_apps/simple-interactive.py

serve

bokeh serve bokeh_apps/simple-interactive.py --show or bokeh serve bokeh_apps/images.py --show