>Every extra variable (e.g. column count) you add is a new opportunity for something to go wrong.
I see that, however you need something to make project that reduces columns cheaper, don't you? On the other hand, if you use (rows, cpu) costing tuple, what would you consider cheaper (2 rows, 3 cpu) or (3 rows, 2 cpu)? The easiest way to compare costs is to use "rows + X*cpu" formula, that basically requires some coefficient. > query failed to use JdbcSort Well, I just picked a first JdbcTest test that did instantiate JdbcSort. Of course we need a separate issue for sorting (i.e. simplest case for it). Vladimir
