Expansion and Reduction

If you come from an imperative background, think of a pipeline as a loop over items. Each step sees items in order and can emit zero, one, or many items for each input.

The four step shapes describe that behavior in plain terms:

One-to-One (1 → 1)

For every input item, produce exactly one output item.

Think: for each item, return one item.

Typical use: validation, enrichment, format conversion.

One-to-Many (1 → N)

For every input item, emit multiple outputs.

Think: for each item, split into many.

Typical use: splitting a batch file into records, expanding a summary into details.

Many-to-One (N → 1)

Consume a stream of items and return a single result.

Think: read many, aggregate into one.

Typical use: rolling up payments into a single report, generating a file.

Many-to-Many (N → N)

Consume a stream and emit a transformed stream.

Think: filter + transform + emit.

Typical use: filtering, joining, or fan-out with transformation.

Why This Matters

• Expansion steps increase the number of items in flight.
• Reduction steps collapse the stream into a final result.
• The shape determines how you place side effects and where you expect backpressure.

If you can reason about a loop and a list, you already understand these shapes.