Row triggers need 2 functionalities to be implemented:
- possibility to fetch a row by completing the base list of Query to get the OLD tuple before a trigger is fired. This can use the RETURNING process to be accomplished.
- possibility to update remotely NEW tuples after the trigger has been fired on the OLD tuple. For the time being the update is only done at local node, what is incorrect as data is locally, fatally on remote Datanodes.