Lee Crain wrote: > > The reason I was wondering about indexing is that the indexes contain > exactly the same data as is in the table and this seemed like an > unnecessary duplication of data. >
Actually this can be beneficial at least speed wise. SQLite has an optimization that will return data directly from the index where possible. This eliminates the extra work of using the rowid in the index entry to locate the row in the table (an O(log N) operation) so that it can retrieve the unindexed column data. If all your data is duplicated in the index, it can always use this optimization, and will in fact never pull data from the table itself for a query that uses the index. The table (along with the index) will only be accessed when rows are inserted, updated, or deleted. Dennis Cote _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
