#[arrow_custom_data]Expand description
Adds Arrow schema, encoding, and decoding implementations to a custom data struct.
Apply this above #[nautilus_model::custom_data] for macro-generated model behavior, or use it
alone when the model traits are implemented manually. Call
nautilus_serialization::ensure_custom_data_registered::<T>() before catalog or streaming
serialization.
Supported field types include InstrumentId, AccountId, Currency, BarType, Params,
Price, Quantity, Decimal, optional prices and quantities, UnixNanos, f64, f32, bool,
String, u64, i64, u32, i32, Vec<f64>, and Vec<u8>.
Prices and quantities use nullable Decimal128(38, 16) columns. Vec<u8> is the deliberate
opaque-byte escape hatch for custom schemas.
Decimal values preserve their numeric value at scale 16, but not the source value’s
trailing-zero scale.
Use #[custom_data_field(native_enum)] on a native enum that implements Display and
FromStr to encode it as a compact dictionary of display names.
Use #[custom_data_field(serde)] on a field to store any Serde serializable field as a
Serde JSON-backed Arrow Utf8 column. Python field access for such fields comes from
nautilus_model’s custom_data macro, not this macro.
Use #[arrow_custom_data(pyo3)] to generate PyArrow encode_record_batch_py and
decode_record_batch_py methods; this is independent from the model macro’s pyo3 option.
Use stub_module = "nautilus_trader.<module>" with pyo3 to emit pyo3-stub-gen metadata.