Clemens,
Considering that scrollable result sets on _any_ RDBMS that has them are
pretty heavy weight and can hold a lot of locks, etc, in my opinion it'd
make more sense to do a SELECT COUNT(*) with the same WHERE clause, and
use that, or re-architect your solution to not need to know the ultimate
amount of rows (which is what a lot of applications do).
Perhaps an alternative is to include the COUNT(*) as one of the columns on
your ResultSet. It would be repeated in every row, of course, but I'm
uncertain as to whether there's a point where the extra weight on the
ResultSet overcomes the extra wait on making a second call.
Michael