On Nov 22, 2013 6:08 AM, "Simon Slavin" <slav...@bigfraud.org> wrote: > > > On 22 Nov 2013, at 1:24pm, Clemens Ladisch <clem...@ladisch.de> wrote: > > > Simon Slavin wrote: > >> All statements are entirely enclosed in a transaction. > > > > No, automatic transactions start with the first sqlite3_step() and end > > with either sqlite3_reset() or sqlite3_finalize(). > > Thanks, Clemens. Interesting, and it lets me tell Sascha to ignore what I posted earlier. >
While our long lived prepared statements are reset many times throughout their lifecycle, there are several of them in use at the same time, and probably never in a state where all of them are reset. The statements feed a UI and are constantly refreshed, requeried, and reset based on user actions. We actually hold a cache of prepared statements so we know which ones are open, maybe as an experiment I could try calling reset on all of them and see if that gives the checkpointer enough time to do its thing? // Sascha _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
