On Mon, Nov 3, 2014 at 6:48 AM, Clemens Ladisch <clem...@ladisch.de> wrote:
> Paul wrote: > > Are additional indices, created for WITHOUT ROWID, potentially less > > efficient and more cumbersome? > > For tables with a rowid, the index stores the indexed columns and the > rowid. For WITHOUT ROWID tables, the index stores the indexed columns > and the primary key. > And, lookup by rowid is (usually) faster than lookup by arbitrary primary key. So, yes, WITHOUT ROWIDs might be a little slower when using secondary indexes. But in a complex system, many factors come into play. It is best to give it a try. Realize that any schema that works with WITHOUT ROWID on a table, will work just as well with the WITHOUT ROWID omitted. So you can test your application both with and without the WITHOUT ROWID clause and see which gives the better performance, without making any changes to your internal queries. -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
