I think I found the problem. I used a version of pysqlite compiled against sqlite3.2.8. When used with the shared library from any sqlite 3.3.x version, every sql operation eats all memory.
Thank you all for the help On 7/14/06, Eduardo <[EMAIL PROTECTED]> wrote:
At 17:43 13/07/2006, you wrote: >Even the smallest sql statement (like PRAGMA temp_store = 1) triggers >this behaviour when using sqlite 3.3.x. I forgot to mention that I >connect to a :memory: database and then attach 2 database to the >connection. Perhaps the 2 database attached are converted to :memory: database also and eat your ram. What size has each one? Try to use the :memory: database as file database (so no :memory: databases). I suggest to don't work with :memory: , use a ramdisk. Here you have a litttle tutorial about doing so http://www.vanemery.com/Linux/Ramdisk/ramdisk.html . Also, as it's a ramdisk you can switch off journaling and you can format it with a filesystem with compression on the fly. As counters, the same as :memory: databases, when a electrical failure happens, data is lost. You can trigger a .dump to file easily at fixed times or use more specialiced ramdisk drivers which do it automatically. ... >-- >Robert Wallner ----------------------------------------------------------------------------------------------------------------------------- La diferencia entre la teoria y la practica es que en teoria no hay, pero en la practica si
-- Robert Wallner