A known issue with the fuzzer is that comparison of query results often takes longer than running the query. We could get more coverage per run out of our fuzzers by porting the result comparison to C++ (assuming this would yield a perf improvement).