Op zondag 19 februari 2006 13:23, schreef Nemanja Corlija: > On 2/19/06, SanjayK <[EMAIL PROTECTED]> wrote: > > I am using a virtual tree control to display columns dynamically when > > needed from the sqlite database. While it works ok for display purposes, > > in certain other operations, I need to improve the speed. I found that > > the new sqlite random access (even with prepared/transaction) approach is > > about 15 times slower than my earlier design where I was using a direct > > access file with read, seek, etc on Windows. > > > > In spite of this, sqlite has several advantages and I am staying with it. > > I am looking for speed improvement suggestions. Somewhere in a thread I > > read "disable indexing." I can't find any reference to how to do it in > > the docs or in this group. How do I disable indexing? I will also > > appreciate any other suggestions for speed improvement too. > > Disabling index might come in handy in some cases. For example, I was > able to speedup one query a lot by disabling index for sorting. For > some reason SQLite 2 was going back to disk to sort by index even > though it had all data needed in result set already. You can disable > the index like this: > SELECT * FROM foo ORDER BY +bar;
I used a lot of the opcode from 2.8.x (forgot exact which one) It did not use an index on sorting, but it sorted the result-set > > where bar is your indexed column. > > If that still doesn't help, you'll have much better chance of getting > some help here on the list if you post more info. Such as SQLite > version used, schema of tables involved in query and query it self. > EXPLAIN's output for that query wouldn't hurt also. If you're using > some wrapper it would be good to run that query from sqlite shell and > see how fast that goes. > > -- > Nemanja Corlija <[EMAIL PROTECTED]> -- Met vriendelijke groet Bert Verhees ROSA Software
