Panel
chirashi.panel
A Docker Compose file for Panel alone, generated from the catalog and checked by the safety inspector. No account needed.
Python framework for building interactive dashboards and applications from analysis code, connecting widgets to plots and tables from the main visualisation libraries without front-end code.
Panel composes widgets, plots, tables and text into layouts and binds widget values to functions, so changing a control re-runs only the dependent computation and updates the affected panel. It renders output from most of the Python plotting ecosystem rather than requiring one library, which lets existing analysis figures move into an application unchanged. The same object can be displayed inline in a notebook or served as a standalone web application from the command line, which shortens the path from exploration to something shareable. Sessions are per user, so state stays isolated between people using the same served application.
You know it worked when
- The application serves and loads in a browser.
- A widget change triggers a visible update to the dependent panel.
- A plot from the intended library renders correctly.
- Two browser sessions maintain independent state.
Known sharp edges
- Each session holds its own state in server memory, so concurrent users multiply memory use and a long-running callback blocks that session's updates.
- The script is executed for every new session, which turns expensive setup work at module level into a per-user cost unless it is cached deliberately.
- Websocket connections require any reverse proxy in front to be configured for them, and the symptom of a missing setting is a page that loads but never updates.
- The allowed websocket origin must include the address users actually reach, or the application renders and then silently refuses to connect.