This ResultSet evaluates grouped, non distinct aggregates. It will scan the entire source result set and calculate the grouped aggregates when scanning the source during the first call to next(). The implementation is capable of computing multiple levels of grouping in a single result set (this is requested using GROUP BY ROLLUP). This implementation has 3 variations, which it chooses according to the following rules: - If the data are guaranteed to arrive already in sorted order, we make a single pass over the data, computing the aggregates in-line as the data are read. - If the statement requests either multiple ROLLUP levels, or a DISTINCT grouping, then the data are first sorted, then we make a single pass over the data as above. - Otherwise, the data are sorted, and a SortObserver is used to compute the aggregations inside the sort, and the results are read back directly from the sorter. Note that, as of the introduction of the ROLLUP support, we no longer ALWAYS compute the aggregates using a SortObserver, which is an arrangement by which the sorter calls back into the aggregates during the sort process each time it consolidates two rows with the same sort key. Using aggregate sort observers is an efficient technique, but it was complex to extend it to the ROLLUP case, so to simplify the code we just have one path for both already-sorted and un-sorted data sources in the ROLLUP case.