On Sat, Jun 23, 2012 at 9:21 PM, Dennis Volodomanov <i...@psunrise.com> wrote: > On 22/06/2012 9:48 AM, Dennis Volodomanov wrote: >> >> I'll see if the new compilation options still make this happen, but it >> takes a couple of hours for each test due to data volume and I'd need to run >> a few tests (unless it occurs right away of course). I'll post back. >> > > This hasn't occurred yet, but I did manage to replicate this in another way > (using the new compilation options) by stopping (in debug)/crashing the > second writer process. It appears (albeit hard to say decisively) that the > entry in the table that was written by the crashed process (and thus it's > wal is not flushed) is the one that comes up in count, but can't be selected > by the other process. Does this make sense or is it just a coincidence that > I'm seeing?
Such thing shouldn't ever happen, otherwise SQLite has a serious bug. Pavel _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users