How should we orgnize Jdbc connectors?

[FANNG1]

1. use a separate connector, like mysql-connector, pg-connector
2. use a common connector, like jdbc-connector with different types like mysql, pg

[FANNG1]

I prefer using a separate connector. because:

1. It's straight-forward to end users, if using jdbc-connector, provider and type identifies a real connector. this produces complex for users and other modules
2. the main benefit of jdbc-connector is code-reuse, different types just provide TypeTransform and ExceptionMapper. If a engine 100% compatible with MySQL, we could just reuse MySQL connector without adding a new connector. I think this condition is rare, every engine has it's own features (column types, table properteis), even if a engine 100% compatible with MySQL, we could simple add a xx-connector which extends JdbcConnector with existing MysqlTypeTransform and MysqlExecptionMapper, it doesn't take much work.
3. Trino, Metacat use separate Jdbc connectors too. if we collaborate with Trino , it's more straight forward to provide a separate jdbc connector.

[FANNG1]

@jerryshao @xunliu @Clearvive @mchades @yuqi1129 what's your advise?

[diqiu50]

I agree case 1.
The JDBC connector is an abstract concept. It has specific implementations like MySQL connector, PostgreSQL connector, and so on. For users, it's easier to understand something like MySQL connector than just a generic JDBC connector. Products like DataGrid, Nativecat, DBeaver, Trino, etc., also enable users to directly use specific implementations like MySQL connector.