> Also it's quite known that > creating index after inserting all rows is much faster than creating > index before that. So it can be even beneficial in inserting huge > amount of rows somewhere in the middle of the work: first delete all > indexes, then insert rows, then create indexes once more. >
Pavel, please add some reference (for example from docs or drh), my tests showed that for large data there almost no difference in speed. And at the first place I thought that the algorithm is really different, now I almost sure that building index from scratch is just enumerating records building the tree. CMIIW Max _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users