Assuming memory is sufficiently inexpensive, I would think that it would almost always be useful to build an index for any field in a join rather than doing a full scan. (Or better yet, build a hash table if memory is sufficient.) Indices maintained in the database then become optimizations to avoid starting the query with an index build.
Mark _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users