Sharing dashboards

Panel has another neat trick: it can create a dashboard that is separate from the whole jupyter notebook interface.

In the previous section, we created a dashboard with a title, some text and a plot:

pn.Column(pn.pane.Markdown("# This is a horizontal barchart"),
          pn.Row(
                 pn.panel("This barchart with 4 categories shows us how we can combine different panes, including vega plots"),
                 pn.pane.Vega(spec)
          ))

This code was embedded in the rest of the notebook, so actually looked like this:

Obviously, this is good enough if we’re still in the developing phase, but at some point, we will want to have a cleaner solution.

Nothing could be simpler: just by add .show() to the code above, jupyter notebook opens a new window with just this dashboard.

pn.Column(pn.pane.Markdown("# This is a horizontal barchart"),
          pn.Row(
                 pn.panel("This barchart with 4 categories shows us how we can combine different panes, including vega plots"),
                 pn.pane.Vega(spec)
          )).show()

This starts a new notebook process on your computer and opens a new browser tab with just the dashboard.

More permanently

In some cases, however, you want to make the dashboard available to a bigger group of people and just want to have it run on some server. In that case we cannot run the notebook interactively and use another solution.

This is a two-step process:

Instead of .show(), call the method .servable() on the dashboard that you want to make available. This will not do anything while you are still in the notebook.
When you have saved the notebook, run panel serve --show my_notebook.ipynb on the command line. This will run the complete jupyter notebook in the backend, take the dashboard that can be identified with the .servable() method, and create a new interface.

This is the output when I run this command:

This output shows us that the dashboard is running on http://localhost:5006/minimal_vega where other people can access it (obviously replace localhost with the correct IP number).

If you have .servable() on more than one dashboard, these will all be included.

Sharing on your own computer

More permanently