On Mon, Aug 17, 2009 at 3:15 AM, Dan Kennedy <danielk1...@gmail.com> wrote:
> >> The INDEXED BY feature was introduced to address concerns that SQLite > >> might > >> suddenly start using a different plan for a query in the field than > >> it > >> did > >> in the office during testing. Either because somebody ran ANALYZE, or > >> because > >> the SQLite version was upgraded. In this situation, some users > >> consider it > >> better to throw an exception than to run the query with a different, > >> possibly > >> slower, plan. > > > > Confusion reigns supreme. Your second last paragraph says (about your > > last scenario) that it uses index i1 instead of the apparently better > > index i2 -- no exception throwing. Your last paragraph indicates > > that in > > this case an exception would be thrown. > > I guess I got that wrong then. Said users considered it better to throw > an error if the index that the author of the SQL query expected it to > use had been removed or radically altered. > > Dan. > Sorry for muddying the waters. I was just trying to make clear that INDEXED BY isn't intended to be used as a tuning mechanism for index selection. My statement should have been clearer. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
