Re: [sqlite] using test_intarray

2010-07-14 Thread Simon Slavin
On 14 Jul 2010, at 8:22pm, Igor Sereda wrote: > I would agree in other cases, however, in our app, the schema is dynamic and > depends on the user data. Normally, we have about 200 tables with two or > three columns each, with indexes almost on each table. Queries with > lots-of-joins are also

Re: [sqlite] using test_intarray

2010-07-14 Thread Igor Sereda
Simon, I would agree in other cases, however, in our app, the schema is dynamic and depends on the user data. Normally, we have about 200 tables with two or three columns each, with indexes almost on each table. Queries with lots-of-joins are also constructed dynamically, and we found SQLite to

Re: [sqlite] using test_intarray

2010-07-14 Thread Simon Slavin
On 14 Jul 2010, at 4:17pm, Igor Sereda wrote: > For example, where we now have a query > > SELECT ...lots-of-joins... WHERE ...lots-of-exprs... AND someColumn IN > (?,?,?,? ...40 params... ) This -- the 'lots-of-joins' and the '40 params' in particular -- suggests you should be rethinking

[sqlite] using test_intarray

2010-07-14 Thread Igor Sereda
Thanks for mentioning test_intarray! I'm now considering rewriting parts of our code because sqlite3_intarray_bind is more powerful than using sequences like ?,?,?... A question: does using a virtual table (or precisely virtual table from test_intarray) affect query optimizer? We have lots of