Question

We have 4 normalized tables, for description sake we will call them t1, t2,t3,t4.

t1 is a parent of t2 which is a parent of t3 which is a parent of t4. These 4 tables store hundreds of columns of data. This data needs to be evaluated for business logic errors. I would like to create an error log for these tables, but storing each error individually will take up too much space because of the large number of permutations that can occur with the error/non-error evaluations of the data.

Would I keep a byte array or something in an error table to signify which error criteria was met or not?

Any other ideas?

Thanks!

Answer 1

If you’re saying there could be t1 * t2 * t3 * t4 records maybe try an error table with a column for t1.id – t4.id, and a pass/fail column for each test.

Something like, (obviously change the column types):

CREATE TABLE error_log
( error_id INT UNSIGNED AUTO_INCREMENT PRIMARY KEY
, t1_id INT
, t2_id INT
, t3_id INT
, t4_id INT
, error_criteria_pass_fail_1 BOOLEAN
, error_criteria_pass_fail_2 BOOLEAN
, error_criteria_pass_fail_3 BOOLEAN
, ...
)

This way you’ll have only one row fore each combination of table id’s and each error test save on that one row.