I am trying to analysis the performance of indexed select. CREATE TABLE t2(a INTEGER, b INTEGER, c TEXT); CREATE INDEX t2c ON t2(c);
I think there may be much more leaf index b-tree pages whose header is '0x0A' if the length of the content of index key 'c' is always 20-25 bytes, as I notice the format of index inside sqlite consist of the index key and rowid. I can establish mapping relation between column 'c' and a new INTEGER column 'd'. Then I am wondering if it is reasonable to create new index t2(d) to get a better performance, as sqlite stores INTEGER in a variable-length way which means there will be less index pages. So if it is correct that the performance of indexed select is up to the number of index pages which is fetched in getPageNormal() within the select? I think it has positive correlation but I do not know if it is the major constraint. And does sqlite have a profile tool to get call tree or execution time of each functions? All I know is VDBE_PROFILE. Thanks for any light you can shed. I want to profile sqlite -- Sent from: http://sqlite.1065341.n5.nabble.com/ _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
