Postgres still doesn't. I usually write a rule to do so. For instance, here's my "ON DUPLICATE KEY UPDATE eid=eid":

  CREATE RULE location_updates_on_duplicate_ignore AS
    ON INSERT TO location_updates
    WHERE (EXISTS (SELECT 1 FROM location_updates WHERE (eid = NEW.eid))) 
    DO INSTEAD NOTHING;

But you can also do much more advanced merge logic by replacing DO INSTEAD NOTHING with DO INSTEAD <statement>. Mine just needed to ignore already-submitted batches.