Question

I have a trigger that stores changes made in a separate table when a user edits the data. This data is written out on a web page beneath the current data in human readable format, i.e.

23/04/09 22:47 James Smith changed Tenant Name from “George Hill” to “George Hilling”.

The trigger I have looks a bit like this – (this is the abridged version).

Two questions:

A) Is this quite costly performance-wise and if so is there a better approach to take?
B) Is there a tidier way to do this without all the IFs, using some sort of loop perhaps?

DELIMITER //
CREATE TRIGGER repair_history AFTER UPDATE ON repairs
FOR EACH ROW
BEGIN
INSERT INTO repair_edit SET repair_id=NEW.repair_id,
edit_date_time=NEW.edit_date_time, edited_by=NEW.edited_by;

IF OLD.tenant_name != NEW.tenant_name THEN
INSERT INTO repair_history SET edit_id=LAST_INSERT_ID(), field='tenant_name',
former_field_value=OLD.tenant_name, new_field_value=NEW.tenant_name;
END IF;

IF OLD.priority != NEW.priority THEN
INSERT INTO repair_history SET edit_id=LAST_INSERT_ID(), field='priority',
former_field_value=OLD.priority, new_field_value=NEW.priority;
END IF;

IF OLD.property_id != NEW.property_id THEN
INSERT INTO repair_history SET edit_id=LAST_INSERT_ID(), field='property_id',
former_field_value=OLD.property_id, new_field_value=NEW.property_id;
END IF;

IF OLD.type_id != NEW.type_id THEN
INSERT INTO repair_history SET edit_id=LAST_INSERT_ID(), field='type_id',
former_field_value=OLD.type_id, new_field_value=NEW.type_id;
END IF;

END; //
DELIMITER ;

Answer 1

A) Is this quite costly performance-wise

Profile it. Only you know what hardware you’re using, and what volume of updates your database has to handle.

and if so is there a better approach to take?

Consider what’s taking the mosst time in your trigger. Is it the comparisons, or the inserts?

B) Is there a tidier way to do this without all the IFs, using some sort of loop perhaps?

Unfortunately, I don’t know of a way to index into the columns of pseudo tables NEW and OLD