On 29 Mar 2013, at 4:57am, Rob Turpin <[email protected]> wrote: > Thanks for looking into this. So I get the notion of explicit scheduling > with lsm_work, but as I currently understand the sqlite4 API, explicit > scheduling can't be done with a sqlite4 database object. Is that correct?
Can you explain what it is that you care about the scheduling of ? You can schedule whatever calls the SQLite API versus any other things your program is doing, but you can't schedule how SQLite works internally. Changes to a SQLite database are performed under the SQL definition of a change: one transaction is one alteration; one transaction can involve creating/destroying/changing any number of rows of any number of tables; a number of transactions must change the database in the order that those transactions are committed. SQLite does these things internally because ACID requires it. I don't know of any reason to mess with SQLite internally since you signal everything you need to it by when you execute "COMMIT". Something about ism_work can be found here: <http://sqlite.org/src4/doc/trunk/www/lsmperf.wiki> but I don't know whether that addresses your problem. Simon. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
