Well… this is embarrassing… Turns out that I'm closing the connection (in some random place) without properly clearing the instance (the object that interfaces to the c api).
I'm hoping this isn't a red herring! Lots of testing tomorrow… I'll report back. Thanks!! -- AJ On Tuesday, 19 March, 2013 at 12:21 AM, Simon Slavin wrote: > > On 19 Mar 2013, at 4:05am, Alykhan Jetha <[email protected] > (mailto:[email protected])> wrote: > > > What's actually happening is I'm getting a SQLITE_NOMEM when I run the > > queries in Process A (after Process B finished and checkpoints). > > Once again the tactic pays off: check every single result code SQLite > returns, even if it's nowhere near your problem statement. > > Are you using _exec() or _prepare(),_step(),_finalize() ? Both _exec() and > _prepare() may unexpectedly use a ton of memory. The other two should stand > very little chance of returning SQLITE_NOMEM. > > SQLite has internal memory debugging code: > > <http://www.sqlite.org/malloc.html> > > but you may be more likely to find your problem using a standard memory tools > like valgrind. > > Simon. > _______________________________________________ > sqlite-users mailing list > [email protected] (mailto:[email protected]) > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
