BTW, this is why I added docid as an alias for rowid for fts3 (rowid-versus-VACUUM was the initial reason for revving fts2 to fts3). Saying "rowid is just like rowid in other tables, except different" seemed more confusing than it was worth.
Though since I'm not writing the new documentation, I'm more than happy to be ignored :-). -scott On Thu, Dec 3, 2009 at 4:28 PM, John Brooks <[email protected]> wrote: > In the new fts3.html, the statement: > > "each FTS3 table has a 'rowid' column that behaves like an INTEGER > PRIMARY KEY, except that values remain unchanged if the database is > rebuilt using the VACUUM command." > > is incorrect; an INTEGER PRIMARY KEY column *does* remain unchanged > during VACUUM, in contrast to a usual ROWID. It is unclear here if the > FTS3 rowid does or does not change during vacuum. > > - John Brooks > > On Thu, Dec 3, 2009 at 4:15 PM, Nicolas Williams > <[email protected]> wrote: >> Use the glob operator. >> _______________________________________________ >> 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 > _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
