On 19 Aug 2010, at 11:18pm, Eric Smith wrote: > The schema is roughly > > create table records(__recno INTEGER PRIMARY KEY, fileId, data); > > So sometimes we would *like* to say "delete from records where > fileId=7". > > But that is bad because does a full table scan. > > So the next cut is to say "create index recordsIdxFileId on > records(fileId)". > > But that is bad because it is a huge burden during INSERT and is not > used often enough (or with enough values) to justify its existence. > > What I really want is to be able to say "create index > recordsIdxFileIdOnFile3 on records(fileId) where fileId=7". > But sqlite doesn't do that.
But in order to create that partial index, SQLite would have to scan the entire table anyway. Otherwise how would it know which records had fileId=7 ? Simon. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
