We currently have a single implementation of the Table and Column metamodel. This has led to the addition of generic solutions to the metamodel to track connector-specific attributes, such as Oracle's OBJECT_ID and OBJECT_DATA_ID attributes.

There are other use cases where a connector may need to track specific metadata for a table or column that isn't required by other connectors, and adding this to the base class implementation has the potential to create unnecessary overhead where it's not used.

For DBZ-8597, this would allow tracking column positions not only based on JDBC metadata but also based on custom calculations such as excluding generated columns from the position details. This would yield a reliable cache solution stored in the Column instance that would work for Oracle and streaming changes for tables with generated columns. This is currently cached independently and moving this directly into the metamodel would make far more sense.