"Brandon, Nicholas \(UK\)" <[EMAIL PROTECTED]> wrote: > > I've noticed this question [making count(*) faster] has been raised > a few times in the past and the workaround proposed is to create a > "count" table to eliminate the need to do a table scan. I was > wondering whether it is worth adding this feature to Sqlite so that > a call to "select count(*)... " (which is not restricted with a > WHERE clause) is retrieved from an internal "count" table? i.e. > this performance workaround is part of the core sqlite code. >
I considered but rejected that idea when originally designing the SQLite file format. Keeping a count involves extra I/O which slows down every INSERT or DELETE operation. I did not want users who do not use count(*) to have to pay that penalty. -- D. Richard Hipp <[EMAIL PROTECTED]> ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
