> Pavel Ivanov wrote: > > It mentions "from" and "column_name" where column name in this > > particular case is "selected". ;-) > > > > @Igor: I thought that sql standard in this case doesn't guarantee > that > > outer select will return rows in the same order that were enforced in > > inner select by "order by", does it? > > Yes, technically, I guess, the order is not guaranteed, though SQLite > is > highly likely to return records in the same order subselect generated > them. Note that Rand Huck's temporary table solution suffers from the > same problem. > > My solution can be modified to give you that extra peace of mind: > > select id, url, selected from ( > SELECT id, url, selected, name FROM db1.test UNION > SELECT id, url, selected, name FROM db2.test > ORDER BY name ASC, id DESC LIMIT 100 > ) > order by name ASC, id DESC; > > Of course, both are solutions in search of a problem. It is likely > trivial to simply ignore the extra column in the host application. > > Igor Tandetnik > >
Many thanks for all the help on this thread. Yes it is trivial to ignore the extra column in the host application. I naively thought there might be some performance gains in not retrieving the name, since it wasn't actually used directly. Perhaps I thought the indexes would be enough to do the sorting. But I now realise that was off track. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users