On Wed, Oct 23, 2013 at 1:50 PM, Vinay Boocha <[email protected]> wrote:
> Dear Rich, > > Thanks for the response. I guessed it might need some cache but thought it > will refetch every time to ensure consistency. > > My main issue is that temp views are not picking up the update after the > pages of joined tables got updated (tables are updated by another actor in > the system). > I don't understand. Manifested views should only last for a single SQL statement then be discarded. Are you claiming that a manifested view is persisting past a single SQL statement? Maybe you can provide more details about your problem? > > Any way to notify the temp views to refetch the pages or always refetch > pages before executing a query? > > My VFS is performing a multi-master replication without the knowledge of > SQLite layer. > > > On Wed, Oct 23, 2013 at 12:25 PM, Richard Hipp <[email protected]> wrote: > > > On Wed, Oct 23, 2013 at 1:20 PM, Vinay Boocha <[email protected]> wrote: > > > > > Dear All, > > > > > > We are observing that temp views are using page cache even after > setting > > > the cache_size= 0. I am stuck debugging this issue for the last two > days. > > > Any pointers will be greatly appreciated. > > > > > > > Depending on the query and the schema, the view might need to be > evaluated > > ("manifested"). Where is that manifested view to be stored, if not in > the > > page cache? > > > > Why is SQLite using memory to help it answer a query an "issue" for you? > > > > > > > > > > When we run a SQL query multiple times on a temp view our custom vfs is > > > getting read call only for offset=24 and length=16. I believe its > > checking > > > if the file is updated or not. > > > > > > > Correct. Just because the database is read-only to you doesn't mean it > is > > read-only to every other process in the system. > > > > > > > > > > > > *Details:* > > > > > > > > > 1. Temp view is created by joining two tables. One of the table is > in > > an > > > attached database. > > > 2. pragma cache_size= 0 is set immediately after opening the > > connection. > > > 3. The cache_size zero is confirmed based on vfs calls for the table > > in > > > the main database. > > > _______________________________________________ > > > sqlite-users mailing list > > > [email protected] > > > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > > > > > > > > > > > -- > > D. Richard Hipp > > [email protected] > > _______________________________________________ > > sqlite-users mailing list > > [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 > -- D. Richard Hipp [email protected] _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
