pt-query-advisor RES.001 is incorrect

[Rob Smith]

The following query incorrectly triggers RES.001

SELECT cola, MAX(colb) FROM table WHERE cola = 123 GROUP BY cola;

All the aggregate functions I've tested seem to result in the same incorrect RES.001 trigger.

[Baron Schwartz (baron-xaprb)]

Actually that test case shows a different thing we ought to be catching: if you GROUP BY something that's constrained to be a constant, you should really not GROUP BY it. We might want to create a new rule for that.

[Rob Smith (kormoc)]

In reality, the WHERE is cola IN (123, 456, 789, etc), in which case the group by does make sense. I just simplified it. Glad my oversimplification found another issue to catch :)

[Brian Fraser (fraserbn)]

The attached branch only fixes the reported bug, but doesn't add the new rule that Baron suggested.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PT-779