I was excited about the option column.propagate.source.type specifically for numeric types, because I cannot find any other way to propagate easily decodable data and still maintain precision/scale. So I want to encode numeric values as strings, but retain scale/precision.

Herein lies my problem: I want to propagate source type from postgres based on the source column data type, not the column name. It would be very difficult for us to configure this by name, because it would necessitate not only finding all numeric columns, but somehow updating the configuration files when a data type changes.

Thoughts?