> On Jan 17, 2018, at 6:08 AM, Dominique Devienne <ddevie...@gmail.com> wrote: > > you gain much faster table scans, because you no longer have to read those > large columns when going from row to row during a scan.
I don’t see how this helps Nick's specific query, though, since it fetches all of the text columns. And with the auto-index on ‘b’, there’s no table scan going on. Nick — you haven’t said what API you’re using to access SQLite. If it’s not the raw C API, there may be overhead involved in that layer, especially dealing with large strings. What I would do is run a CPU profiler during the query and see if there are any hot-spots you can do something about. (In my opinion, profiling is greatly underused. I keep coming across people complaining about performance or asking how to speed something up, and usually they haven’t even considered profiling. Maybe I’m spoiled because in the Mac/iOS environment both the Instruments app and the ‘sample’ tool are easy to use, but even if the profiler is tricky to learn, it’s a worthwhile investment.) —Jens _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
