Unpack Record

Splits a record-typed stream into one output stream per field, preserving declared field order. Inverse of {{component:pack_record}}.

How it fits

{{type:{a:a,b:b,c:c}}} -> {{component:unpack_record}} -+-> {{type:a}}
                                                   +-> {{type:b}}
                                                   +-> {{type:c}}

Typical backends

• Per-field analytics: record-emitting source -> {{component:unpack_record}} -> per-field transforms -> downstream.
• Edit one field, then repack: typed source -> {{component:unpack_named}} -> {{component:unpack_record}} -> transform -> {{component:pack_record}} -> {{component:pack_named}}.
• Detection split: detection record -> {{component:unpack_record}} -> separate box / class / score streams.
• Frame metadata extraction: frame record -> {{component:unpack_record}} -> per-field {{component:output_json_http}}.

Caveats

• All output streams tick together at the input cadence; consumers that do not need every field still pay the synchronisation cost.
• Field order is fixed by the record schema; reordering downstream must use explicit wiring.
• Purely a type-level split — no runtime copy beyond the per-field demux.