Feature request: Show data type in table output

[amoeba]

Currently, when I run a query, I get something like,

> select 1;
┌─────┐
│ 1   │
├─────┤
│ 1.0 │
└─────┘

I think it'd be an nice improvement of the output showed the data type, similar to DuckDB:

D select 1;
┌───────┐
│   1   │
│ int32 │
├───────┤
│   1   │
└───────┘

[esadek]

I'm open to having this added as an optional feature. Not sure how useful it would be for most use cases since it would show Arrow schema types instead of the database schema types.

[amoeba]

An option would work too. I think such a feature could still be useful to help disambiguate Arrow types that print identically, for example,

> SELECT
.       -- Numeric types
.       CAST(1 AS TINYINT) as tinyint_col,
.       CAST(100 AS SMALLINT) as smallint_col,
.       CAST(10000 AS INTEGER) as integer_col,
.       CAST(1000000000 AS BIGINT) as bigint_col,
.       CAST(12345678901234567890 AS HUGEINT) as hugeint_col,
.       CAST(123456789012345678901234567890 AS UHUGEINT) as uhugeint_col,
.       CAST(3.14 AS FLOAT) as float_col,
.       CAST(2.718281828 AS DOUBLE) as double_col,
.       CAST(99.99 AS DECIMAL(10,2)) as decimal_col;
┌─────────────┬─────────────┬─────────────┬────────────┬─────────────┬─────────────┬───────────┬─────────────┬────────────┐
│ tinyint_col │ smallint_co │ integer_col │ bigint_col │ hugeint_col │ uhugeint_co │ float_col │ double_col  │ decimal_co │
│             │ l           │             │            │             │ l           │           │             │ l          │
├─────────────┼─────────────┼─────────────┼────────────┼─────────────┼─────────────┼───────────┼─────────────┼────────────┤
│ 1           │ 100         │ 10000       │ 1000000000 │ 12345678901 │ d20a3f4eeee │ 3.14      │ 2.718281828 │ 99.99      │
│             │             │             │            │ 234567890   │ 073c3f60fe9 │           │             │            │
│             │             │             │            │             │ 8e01000000  │           │             │            │
└─────────────┴─────────────┴─────────────┴────────────┴─────────────┴─────────────┴───────────┴─────────────┴────────────┘

As a separate feature idea, implementing a DESCRIBE directive that doesn't just go down to the underlying database would be cool too.

[lidavidm]

I'm working on having drivers show the type name in column metadata too: adbc-drivers/validation#142

(Albeit the format that was pushed for makes it harder to do this generically.)