panel-vegafusion 0.0.1

Build interactive big data apps with Altair and Vega easily using Panel + VegaFusion.

Panel VegaFusion

WORK IN PROGRESS. PROOF OF CONCEPT WORKING. PACKAGE NOT WORKING!

The Panel VegaFusion pane allows you to create interactive big data apps based on the Altair plotting library and the Vega visualization specification.

It is all powered by VegaFusion which provides serverside acceleration for the Vega visualization grammar.

Reference Example

import altair as alt
import panel as pn
from panel_vegafusion import VegaFusion
from panel_vegafusion.utils import get_plot, ALTAIR_BLUE, get_theme

pn.extension(template="fast")

theme=get_theme()
alt.themes.enable(theme)

plot=get_plot() # Can be replaced any Altair plot or Vega Specification

component = VegaFusion(plot, height=800).servable()

pn.state.template.param.update(
    site="Panel VegaFusion", title="Interactive BIG DATA apps with CROSSFILTERING for Altair and Vega",
    accent_base_color=ALTAIR_BLUE, header_background=ALTAIR_BLUE,
)

Todo

This needs to be done before alpha release

• [] Fix all errors in invoke test.all
• [] Make python package installable (and release it)
• [] Implement way to get rid of the user having to serve the assets manually via --static dist=./src-js/dist.
• [] Get things working on Binder

License - AGPLv3 - IMPORTANT

This Panel Vegafusion project is AGPLv3 Licensed because VegaFusion is AGPLv3 licensed and requires the author to provide this application's source code upon request.

SO PLEASE INVESTIGATE THE LEGAL ASPECTS ON YOUR OWN. YOU WILL BE USING THIS PROJECT AT YOUR OWN RISK ANYWAYS!

References

• VegaFusion
• VegaFusion documentation
• Panel
• Panel Discourse
• Awesome Panel
• DataShader (A mature alternative for big data visualization)
• hvplot/ HoloViews (Provides alternative, easy to use crossfiltering using familar Pandas .plot api)

Issues Identified

• vegafusion/vegafusion #64 - Altair Dark theme not working
• Bokeh Discourse - Cannot bokeh build extension with wasm dependency
• bokeh/ipywidgets_bokeh #46 - Not working with VegaFusionWidget
• holoviz/param #597 - Add edit_readonly
• holoviz/panel #3149 - Object of type Chart is not JSON serializable
• vegafusion/vegafusion #62 - Please support Panel
• vegafusion/vegafusion #63 - Please provide simple .js build

Develop

Install

git clone https://github.com/MarcSkovMadsen/panel-vegafusion.git
conda create -n panel_vegafusion -c conda-forge python=3.9 nodejs
conda activate panel_vegafusion
pip install -e .[all]
cd src-js
npm install --save-dev webpack-cli
npm install
cd ..

Build

Javascript package

invoke build.js

Python package

invoke build.package

Test

pytest test.all

Serve Dev App

For now you can serve an example with hot reload via

panel serve 'tests/apps/test_dev_app.py' --autoreload --show --static dist=./src-js/dist

Release Python Package

Before releasing please make sures you have

• updated all version numbers
• build all packages
• run all tests with succes

python -m twine upload dist/*<VERSION>*

to deploy the package 📦.

If you want to upload to Test Pypi first you can do so by adding --repository testpypi.

Reference

You can also find inspiration in the original Jupyter VegaFusion reference example via

jupyter lab tests/reference_example.ipynb