Richard Hipp <drh@...> writes: > The combination of schema, sqlite_stat1, and query is normally sufficient, > yes. > > Can you change (and shorten) the table and index and column names to > obscure their meaning and origin, and give us written permission to include > your case in the published test suite for SQLite? That would be even > better.
Done; please find the sample database here: www.enzinger.net/archives/ngqptest.zip No problem with using it in the published test suite. The query in question is in VIEW ngqp_test. As said, it demonstrates a dramatic performance regression since version 3.7.15 whenever sqlite_stat1 is absent. Thanks, Wolfgang _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
