One more thing: On Sun, Jun 3, 2018 at 2:16 PM R Smith <ryansmit...@gmail.com> wrote:
> > - If it is an Index, /and/ the Key repeats magnificently much (Imagine > adding an "Age" column to a phone-book and then filtering on Age, there > will be thousands of people who are all 34, for instance) then you are > better off extracting the set of records to a TEMP table and then > paginating through the temp table's complete dataset and COUNT(*) its > rows. This will be extremely fast, especially if the DB is otherwise > quite hefty, and will allow using the new table's rowids (invisible in > the query) as pagination pegs. Be sure to use a memory-oriented journal > mode and cache settings for this, or if not possible, perhaps even a > second attached in-memory or memory-oriented DB. I'd be worried about memory usage, and also having to keep the temp table between queries until it's no longer needed… I can't predict when the client is no longer interested in the query. If the query would match 100k records, won't they all be copied? _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
