Details
Description
As a user I want to be able to provide my own RowMapper<Tuple> implementation to enrich the jdbc data.
My use case requires me to add timestamp field and a delete flag field to records before they get written to HDFS. To do it, I have to implement a ItemReaderFactory and perhaps extend NameColumnJdbcItemReader. This is to override the afterPropertySet method to change the default implementation.
Otherwise I have to write my own Processor that can add these fields to Tuples, and since tuples are immutable I would have to recreate the tuples with additional fields in the processor. For large load this could be big overhead.
I would love to know any other technique to implement such a use case.