Union
Stack the rows of 2 or more inputs into one, matched by field name (or record position), with the fields reconciled.
What it does
Stacks the rows of 2 or more inputs into one output, the vertical counterpart to a join. Columns are reconciled by field NAME (the default) or by record POSITION; a column that appears in only some inputs is null-filled for the rest. Where the same column has different types across inputs they're promoted to a common type (a whole number and a decimal become a decimal; anything incompatible becomes text). Rows keep their input order: all of input 1, then input 2, and so on. Every row is kept; to drop duplicates, add a Unique node.
Inputs and outputs
| Inputs | As many as you set on the node, 2 to 16. |
| Outputs | One. |
When to use it
- Combine same-shaped data from several sources into one stream (monthly files, per-region exports)
- Append this period's rows onto last period's before loading them once
- Bring branches back together, such as a Join's matched rows plus its Left/Right orphans
Examples
Stack by field name
orders_us + orders_eu → one orders stream
Columns line up by name whatever their order; a column missing from one input is null-filled.
Only the shared columns
When fields differ → Only fields present in every input
Drops any column that isn't in all inputs, so every output row is fully populated.
Stack by position
Column 1 with column 1, column 2 with column 2…
Names can differ; the output uses the first input's names. Differing widths are padded or trimmed.
Tips
- Set the number of inputs (2–16) with the − / + stepper, then connect every input you add; the canvas grows a handle per input.
- "When fields differ" sets the shape: keep all fields (null-filling gaps), keep only the fields every input shares, or error if they don't match.
- Field-name matching is case-exact by default; turn on "Match field names ignoring case" to stack ID with id (the first spelling seen wins).
- Mismatched types are reconciled automatically; use the output grid's Type control to force a specific one.
- To combine columns that have DIFFERENT names, rename them the same upstream (Manage Columns) first, then union by name. Put a Unique node after the union to drop duplicate rows.