On 4/1/19, Jinho Jung <jinho...@usc.edu> wrote: > Hello, > > We are developing a tool called sqlfuzz for automatically finding > performance regressions in SQLite. sqlfuzz performs mutational fuzzing to > generate SQL queries that take more time to execute on the latest version > of SQLite compared to prior versions. We hope that these queries would help > further increase the utility of the regression test suite.
Thanks for the report. Since there are already a bazillion fuzzers for SQLite, may I suggest that you choose a more a more specific and descriptive name for your fuzzer? Perhaps "sql-perf-fuzz" or something similar - so that we know that your fuzzer is targeting performance regressions? > > We are sharing four SQL queries that exhibit regressions in this report. > Here’s an illustrative query: I only got 3 SQL queries. What am I missing? Also, I got bisect results for all three problem that are different from the results you report. When I run bisect, I get the same result for all three test cases: https://sqlite.org/src/timeline?bid=y736b53f57fnbd49a8271dycb1511065dy6c6fb1c6eay30f08d5888y507c43537fy2c8769c69fn4371a0c46en4d0a949fd9n79c073878dy93386a7c97nd840e9bb02 So it was apparently a bug-fix that caused the performance decrease. I have not looked into the details yet. Perhaps there is an alternative fix for the bug that does not cause unnecessary performance loss. -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
