join

join enriches one table with another. The left input is the anchor, especially for how: left; the right input supplies matching columns.

Use one join for one relationship. If the explanation has to say “and then it also joins…”, chain another join so the row-count effect stays inspectable.

Join contract

Field	Required	Notes
inputs	yes	Exactly two upstream tables: left first, right second.
leftKey	yes	Bare column name or expression evaluated on the left table.
rightKey	yes	Bare column name or expression evaluated on the right table.
how	no	inner by default; left keeps all left rows.

Before you join

• Choose inner when unmatched rows should disappear. Choose left when the left table defines the cohort.
• Watch many-to-many relationships. Rime allows them, but they create one output row for every matching pair.
• If keys need normalization, an upstream derive node often makes the matching logic easier to review than expression keys inside the join.

Result shape

default is the joined table. Right-side column names are suffixed when needed to avoid collisions.

The editor/report should make left-vs-right order and row-count expansion visible.

Example

- id: enriched
  kind: join
  inputs: [orders, customers]    # length 2
  leftKey: customer_id
  rightKey: id
  how: left                       # inner | left, default inner

Related

• derive - normalize keys before joining
• Expression language - expression join keys