December 20, 2025
Widget Wars ignite
All about Parameters and Widgets in Databricks Workflows
Databricks Parameters Guide Sparks a Widget War
TLDR: A Databricks how-to shows using widgets to pass parameters—and the `removeAll` cleanup trick. Commenters split between loving widgets for quick fixes and insisting on real job parameters and pipelines, debating reliability, costs, and vendor lock-in—because stable, cheaper data runs matter for teams at scale.
A calm tutorial on Databricks “parameters and widgets” just detonated a full-on Widget War in the comments. The article explains how to pass inputs into notebooks with little on-screen boxes (widgets), how to clean them up with removeAll, and why breaking one giant notebook into smaller ones helps. Then the crowd arrived. The pro-widget camp cheered the simplicity: “flip a dropdown, rerun, go home.” They loved seeing real examples and the reminder that you can pass values between notebooks and Workflows. Meanwhile, the skeptics swarmed in: “Widgets are cute for demos—use real job parameters and pipelines,” they snapped, pointing to Workflows and config files, not clicky UI bits. The hottest complaint? The rename-a-widget-and-it-duplicates quirk and the ritual chant of dbutils.widgets.removeAll()—memed as “the rain dance.” YAML purists rolled in with “put it in code, not buttons,” while fans argued widgets are perfect for analysts and on-call heroes. Cost hawks warned that “click-and-rerun” can waste cloud money; vendor-lock-in doomers shouted “Airflow or Dagster forever!” For the rest of us, the vibe is clear: widgets are handy training wheels, but when it’s production, many want fewer buttons and more discipline. Read the guide, then choose your side—and your widget wand docs.
Key Points
- •Databricks parameters customize Notebook/job execution and can be passed between Notebooks and within Workflows.
- •Parameters can be provided via Databricks widgets or command-line options in tasks.
- •Widgets are created and read using dbutils.widgets APIs (text, get) and can be cleared with removeAll.
- •Widget UI appears only after running the creation code; renaming without cleanup can create duplicates.
- •Supported widget types include text, dropdown, combobox, and multiselect; modular Notebooks are recommended for maintainability.