rfc: Infer Schema

[Xuanwo]

Signed-off-by: Xuanwo github@xuanwo.io

I hereby agree to the terms of the CLA available at: https://databend.rs/dev/policies/cla/

Summary

RFC: Infer Schema

preview of RFC: https://databend-git-fork-xuanwo-inferschema-databend.vercel.app/doc/contributing/rfcs/infer-schema

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Updated
databend	✅ Ready (Inspect)	Visit Preview	Nov 5, 2022 at 5:25AM (UTC)

[sundy-li]

we already support create table as

MySQL [(none)]> drop table a;
Query OK, 0 rows affected (0.004 sec)

MySQL [(none)]> create table a as select 3 as a, '3' as b;
Query OK, 0 rows affected (0.006 sec)

MySQL [(none)]> desc a;
+-------+------------------+------+---------+-------+
| Field | Type             | Null | Default | Extra |
+-------+------------------+------+---------+-------+
| a     | TINYINT UNSIGNED | NO   | 0       |       |
| b     | VARCHAR          | NO   |         |       |
+-------+------------------+------+---------+-------+
2 rows in set (0.003 sec)

[Xuanwo]

#8645 (comment)

I think create table as is different too.

[sundy-li]

looks like it's a table function like select * from infer($location, $file_format)

[Xuanwo]

infer is much more complex that (location, file_format).

We need to accept stage/location like in COPY, file format options and so so. I feel it's hard to implement as a function with named args.

[sundy-li]

I think it's more like a table function.

[Xuanwo]

Is there an existing example about table function in databend?

[sundy-li]

Not yet, but it's in ClickHouse https://clickhouse.com/docs/en/sql-reference/table-functions/s3/

[Xuanwo]

I will take this a look.

databend/src/query/service/src/table_functions/table_function_factory.rs

Lines 65 to 69 in 0847a31

	pub struct TableFunctionFactory {
	creators: TableFunctionCreators,
	}
	impl TableFunctionFactory {