Livedocs treats your notebook as a DAG (Directed Acyclic Graph), meaning it understands dependencies between cells. If Cell A produces a variable used by Cell B, Livedocs knows to run Cell A first, even if B appears earlier in the document.
When running all cells from the main run button (top left), Livedocs uses these dependency chains to parallelize execution where possible, making runs faster and more efficient.
Execution Strategies (per-cell)
From the dropdown next to the cell’s run button:
- Run without cache – Ignore cached results and fetch fresh data.
- Auto – Runs only relevant dependent cells.
- Cell only – Runs just this cell.
- All – Runs all cells in the document.
- Downstream – Runs this cell and all cells depending on it.
Example: If a SQL cell outputs sales_df and a chart cell uses sales_df, running the chart with Downstream will ensure the SQL cell runs first.
Stale Results (Dirty Marking)
Livedocs marks cells as dirty if their inputs or dependencies change. If a cell’s code/data hasn’t changed and it has cached results, Livedocs uses the cache for faster execution.
- Reduces load on warehouses like BigQuery, Snowflake, etc., lowering costs (since they often charge per query).
- Prevents unnecessary re-runs of unchanged cells.
- Improves overall notebook run times.
You can disable caching entirely in report settings or manually choose Run without cache when needed.