[Question] how to convert UBIGINT[] to Vec<u64>

[oliverdding]

I'm writting a little demo with a UBIGINT[] type column in my table.

While trying to query the table, the rustc told me that the trait bound `Vec<u64>: FromSql` is not satisfied:

  let mut stmt = conn
        .prepare("SELECT * FROM users WHERE id = ?")
        .map_err(internal_error)?;

    stmt.query_row(params![id.0], |row| {
        Ok(Response::User(UserResponse {
            arr: row.get(1)?,
        }))
    })
    .map_err(|err| match err {
        duckdb::Error::QueryReturnedNoRows => (
            StatusCode::NOT_FOUND,
            ErrorResponse {
                code: -1,
                reason: "User not found".to_string(),
            },
        ),
        _ => internal_error(err),
    })

Here is UserResponse definition:

#[derive(Debug, Serialize, Deserialize)]
pub struct UserResponse {
    pub arr: Vec<u64>,
}

Here is my table DDL:

    CREATE SEQUENCE IF NOT EXISTS seq_user_id START 0;
    
    CREATE TABLE IF NOT EXISTS users (
        id UBIGINT PRIMARY KEY DEFAULT nextval('seq_user_id'),
        arr UBIGINT[]
    );

---

After browsing the source code, I found that only Vec<u8> type are implementated with FromSql trait

Any advice are appreciated!

[dvic]

If I'm not mistaken, it's not yet implemented:

duckdb-rs/src/row.rs

Lines 545 to 580 in 9af377a

	// TODO: support more data types
	// DataType::Interval(unit) => match unit {
	// IntervalUnit::DayTime => {
	// make_string_interval_day_time!(column, row)
	// }
	// IntervalUnit::YearMonth => {
	// make_string_interval_year_month!(column, row)
	// }
	// },
	// DataType::List(_) => make_string_from_list!(column, row),
	// DataType::Dictionary(index_type, _value_type) => match **index_type {
	// DataType::Int8 => dict_array_value_to_string::<Int8Type>(column, row),
	// DataType::Int16 => dict_array_value_to_string::<Int16Type>(column, row),
	// DataType::Int32 => dict_array_value_to_string::<Int32Type>(column, row),
	// DataType::Int64 => dict_array_value_to_string::<Int64Type>(column, row),
	// DataType::UInt8 => dict_array_value_to_string::<UInt8Type>(column, row),
	// DataType::UInt16 => dict_array_value_to_string::<UInt16Type>(column, row),
	// DataType::UInt32 => dict_array_value_to_string::<UInt32Type>(column, row),
	// DataType::UInt64 => dict_array_value_to_string::<UInt64Type>(column, row),
	// _ => Err(ArrowError::InvalidArgumentError(format!(
	// "Pretty printing not supported for {:?} due to index type",
	// column.data_type()
	// ))),
	// },
	// NOTE: DataTypes not supported by duckdb
	// DataType::Date64 => make_string_date!(array::Date64Array, column, row),
	// DataType::Time32(unit) if *unit == TimeUnit::Second => {
	// make_string_time!(array::Time32SecondArray, column, row)
	// }
	// DataType::Time32(unit) if *unit == TimeUnit::Millisecond => {
	// make_string_time!(array::Time32MillisecondArray, column, row)
	// }
	// DataType::Time64(unit) if *unit == TimeUnit::Nanosecond => {
	// make_string_time!(array::Time64NanosecondArray, column, row)
	// }

[oliverdding]

Hi @dvic , thanks for your reference. I would close this issue and keep an eye on this.