Unpack record

Extracts each field from a record and emits it as a separate output stream, preserving field order.

Use this when

• Downstream components operate on individual fields rather than a full record.
• You need to split a detection or inference result into its components for separate processing paths.
• You want per-field routing without a custom extractor component.

What it does

Consumes one record per tick and emits each field on its own output stream. Input {class: DetectedClass, rectangle: Rectangle<Double>, counter: UInt64} produces three streams: DetectedClass, Rectangle<Double>, UInt64.

Inputs

Type	Description
{$ns:$ts...}	A record value.

Outputs

Type	Description
$ts...	One output stream per field, in declaration order.

Parameters

This transformation takes no parameters.

Works best with

• Pack record — natural inverse.
• Downstream components each specialized for one of the record's fields.

Caveats

• Field order in the output matches the record's declaration order, not alphabetical.
• All downstream streams must be consumed at the same cadence, or back-pressure builds up.