- In the new Genny benchmarks for complex predicates include complex mixed expressions where some of the terms are large $in predicates.
- There should be large $or with same-path equalities that can be reduced to $in.
- Mixes of $or and $in with duplicates so that the $or equalities are subsumed by the $in-s.
- Among the DNF/CNF queries with multiple $in-s there should be cases with
- large $in-s with a lot of duplicates,
- $in-s that can be optimized away due to other predicates, e.g.
(a < 100) AND (a IN [200, 250, 300]) - The idea with both cases above is that in order to truly reduce the query, it is necessary to know the contents of the $in, which will be problematic if the $in is parameterized.
- is depended on by
-
SERVER-84401 Collscan benchmarks - complex $in queries
-
- Closed
-