Simon Slavin wrote: > On 26 Mar 2014, at 1:20pm, Clemens Ladisch <[email protected]> wrote: >> In this case, it appears that listing the rowid explicitly confuses >> SQLite somehow. > > That would be a bug, then ? > > Well, not a bug leading to incorrect results, but certainly one that slows > down searches.
This is not a regression from the pre-NG query planner, and indexes with a duplicated rowid are a little bit more inefficient in the first place, so this might be a case that has not been considered, or not seen worthwhile optimizing for. Regards, Clemens _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
