benbotektips
MIS
My dept. is migrating from 6.0 to 7.1, so I have to test the reports to ensure no compatibility problems. Part of the test is timing how long the report runs under different user profiles in the catalog. There are some reports that pull in data from 2 tables that have no join, thus a cross-product. The solution for 7.1 migration will be eventually to merge the table that has the cross-product fields into another table, thus no more cross-product. But, this will happen in a month, & I must test now. This brings up a dilemma:
The version 6.0 catalog, in the governor section, says don't allow cross products for any profiles, yet the reports all work fine, no cross-product warnings. The 7.1 catalog immediately flagged the report & catalog, and the only way to continue testing was to allow cross-product for the user profiles I'm running. There are some long run times on these, longer than on 6.0. Is this test simply not valid, because in version 6.0, for whatever reason, Impromptu doesn't consider the sql a cross product even though it really is, and the reports return within a reasonable time, but in 7.0, it is a cross-product, and therefore the runtimes are 'artificially' higher than when we rollout to production, with the new merged table? In effect, I'm comparing apples & oranges?
Or, is the test valid, comparing runtimes of 6.0 vs. 7.1, since the generated sql for 6.0 was a true cross-product even if the Impromptu governor didn't catch it, and the runtime is higher for another reason?
In summary, am I testing cross-product queries in 6.0 and 7.1, or am I testing non-cross in 6.0, vs. cross in 7.1?
The version 6.0 catalog, in the governor section, says don't allow cross products for any profiles, yet the reports all work fine, no cross-product warnings. The 7.1 catalog immediately flagged the report & catalog, and the only way to continue testing was to allow cross-product for the user profiles I'm running. There are some long run times on these, longer than on 6.0. Is this test simply not valid, because in version 6.0, for whatever reason, Impromptu doesn't consider the sql a cross product even though it really is, and the reports return within a reasonable time, but in 7.0, it is a cross-product, and therefore the runtimes are 'artificially' higher than when we rollout to production, with the new merged table? In effect, I'm comparing apples & oranges?
Or, is the test valid, comparing runtimes of 6.0 vs. 7.1, since the generated sql for 6.0 was a true cross-product even if the Impromptu governor didn't catch it, and the runtime is higher for another reason?
In summary, am I testing cross-product queries in 6.0 and 7.1, or am I testing non-cross in 6.0, vs. cross in 7.1?