range-scan

Iterate over a key range or index range, producing a stream of rows.

Rust API

// sync_api
pub fn mudu_range(
    session_id: OID,
    start_key: &[u8],
    end_key: &[u8],
) -> RS<Vec<(Vec<u8>, Vec<u8>)>> {
    /* ... */
}

// async_api
pub async fn mudu_range(
    session_id: OID,
    start_key: &[u8],
    end_key: &[u8],
) -> RS<Vec<(Vec<u8>, Vec<u8>)>> {
    /* ... */
}

Conceptual signature

range_scan(
  table: string,
  start_key: KeyExpr,
  end_key: KeyExpr,
  options?: { projection?: [string], limit?: int, for_update?: bool }
) -> [Row]

Parameters

• table — target table or logical collection name.

• start_key — inclusive lower bound of the scan.

• end_key — exclusive upper bound of the scan.

• options.projection — optional list of column names to return.

• options.limit — optional maximum number of rows to return.

• options.for_update — when true, acquires locks on returned rows.

Return value

• A list of (key, value) pairs or Row objects matching the range.

Errors

• InvalidKey: the key expression does not match the table schema.

• LockConflict: for_update is requested and lock acquisition fails.

Transaction semantics

• Range scans honor the current transaction’s snapshot/isolation rules.

• Outside a transaction, visibility follows host default semantics.

Example

let rows = mudu_range(
    oid,
    b"accounts:100",
    b"accounts:200",
)?;
for (k, v) in rows {
    // decode k/v according to the table schema
}

Notes

• Long-running scans should be paginated or limited to avoid holding resources.

• When scanning a partitioned table, the kernel may route the request across workers.