Dataset Registration¶

Dataset registration turns repository captures and sidecars into typed dataset identity. Register a dataset here when multiple commands, receiver runs, tests, or validation workflows need the same capture to resolve with the same metadata and provenance.

Registration Flow¶

flowchart LR
    capture["capture data"]
    sidecar["reviewed sidecar<br/>metadata provenance"]
    registry["dataset registry entry"]
    validation["infra validation"]
    consumers["command receiver<br/>validation workflows"]

    capture --> registry
    sidecar --> registry --> validation --> consumers

Registration Contract¶

input	required decision	owned by
dataset identity	stable name, capture role, and lookup path	infra registry
raw-IQ metadata	sample rate, format, quantization, signal hints, and sidecar source	infra and signal contracts
coordinates	one repository parse rule for ECEF or related coordinates	infra parsing
provenance	reviewed source, front-end context, or notes needed for replay	infra provenance
validation	registry and sidecar checks before higher workflows rely on it	infra validation

Change Rules¶

Register the dataset only when the capture is meant to be reused.
Keep metadata explicit enough that a later receiver run does not need command memory to explain the input.
Do not encode receiver lock policy, signal model assumptions, or navigation science in the registry entry.
Update validation expectations when a sidecar field becomes required.

First Proof Check¶

Inspect crates/bijux-gnss-infra/docs/DATASETS.md, crates/bijux-gnss-infra/src/datasets/registry.rs, crates/bijux-gnss-infra/src/datasets/raw_iq_metadata.rs, crates/bijux-gnss-infra/src/parse/coordinates.rs, and the most relevant dataset parsing, sidecar, or registry tests.