Strategy for rejecting rows

Question

Hi,
I have a problem that would like to bring up for discussion. I have a solution in mind, but would like to listen to other suggestions.
Product   Type   Test1   Test2   Test3   Test4   Test5
YTP1       A        1.0       2.0      3.3
YTP1       A        8.0       9.0      7.0
YTP1       B        5.0       6.0      7.0
BBP1      C         2.1      4.7       6.7
BBP2      A        4.5       4.7      6.8The problem statement is:
Suppose the table A above shows the testing results of a particular Product,Type group. Test1, Test2, Test3, Test4...Test10 are the test attributes. A product,type group is said to pass the test if Test1, Test2, Test3, ...Test10 all satisfied a certain criteria.
The criteria for Test1, Test2,...,Test10 are found in 10 seperate table for a Product,Type group. So, the tables look like this:
For Test1-Test10 look like this:
Product   Type   Pass_criteria
YTP1       A       2
YTP1       B       3
BBP1      C       4
BBP2      A       5So, let's assume that pass when the test attribute is less than the pass_criteria. (this criteria may change to less than, greater than etc)
What are the ways to scan the table A(1 million rows) and get those rows that pass?

Answer 1

Would doing this in SQL faster than say, in pl/sql using row by row checking?PL/SQL is usually slower than SQL. If the vast majority of rows are likely to fall at the first hurdle then possibly the time you would save by avoiding subsequent lookups might be quicker.
Which suggests an alternative approach:
SELECT product, type
FROM
    ( SELECT product, type FROM tablea a, test1 t1
     WHERE a.product = t1.product and
           a.type = t1.type and
           a.test1 < t1.pass_criteria
      INTERSECT
      SELECT product, type FROM tablea a, test2 t2
           WHERE a.product = t2.product and
                 a.type = t2.type and
           a.test2 < t2.pass_criteria
     INTERSECT
      SELECT product, type FROM tablea a, test3 t3
           WHERE a.product = t3.product and
                 a.type = t3.type and
           a.test3 < t3.pass_criteria
     INTERSECT     
     INTERSECT
           SELECT product, type FROM tablea a, test10 t10
                WHERE a.product = t10.product and
                      a.type = t10.type and
                a.test10 < t10.pass_criteria
)With cases like these the easiest way of saying whiich is the faster approach comes down to trace and benchmarking. I wouldn't like to say the above is faster than John's suggestion; I hesitate to bet agaibnst JWS.
Cheers, APC