Preparation node
Manage Columns
Keep, drop, reorder, rename, and retype columns.
What it does
The one place to sort out a table's shape. Keep or drop columns, reorder them, rename them, change their type and size, and leave a description, all in one node instead of four. The row order in the node becomes the column order in the output. A freshly dropped node changes nothing until you touch a column.
Inputs and outputs
| Inputs | One. |
| Outputs | One. |
When to use it
- Drop columns you don't need before writing somewhere
- Rename or reorder for a clean, presentable output
- Change a column's type: text to a number, or a number to text to keep leading zeros
Examples
Keep a few, drop the rest
Turn off “Keep unlisted columns”, then tick only id + name
The fastest way to go from 60 columns to 2.
Rename
cust_id → customer_id
The output carries the new name; nothing upstream changes.
Retype and size
zip → Text, max length 5
Text with a length TRUNCATES anything longer, so use it deliberately.
Exact decimals
amount → Fixed decimal (12, 2)
Precision then scale: the way to stop money being stored as a float.
Reorder
Move id to the top
Row order in this node is the output column order.
Tips
- “Keep unlisted columns” passes through anything you didn't list. Turn it off to output only what you ticked. That is the difference between tidying and selecting.
- A column you explicitly dropped stays dropped, even with “Keep unlisted columns” on.
- A type change that can't fit a value produces null rather than failing the run, so check the Data tab after a Test run when you retype something.
- A column the node keeps or renames that is missing from the input is refused by name, in the builder and at run time, so a rule left over from a source that has since changed cannot slip through. Dynamic Select is the node that tolerates columns coming and going.
- This is the node to reach for before a database load: renaming and retyping here is cheaper than fighting the target's DDL.