Rows in this RowSet are ordered by OrderingList (which is a sequence of (column,ASC/DESC) pairs) The semantics vary a bit across UDOs: For extractors and processors, each vertex output rows in OrderingList order, since there is a single invocation of the processor/extractor per vertex For reducers and combiners, the vertex calls the UDO multiple times, once per group of ON columns. Stream-based reducers/combiners sort the inputs on the ON columns. If those are pass-through, the output is sorted by the ON columns In addition to that sort, we attach the ordering list. Example: REDUCE A ON a, b USING SampleReducer Suppose that SampleReducer has "a" as a pass-through column and declares "c,a" as an OrderlingList Also, assume that the stream-based reducer sorts the input on "a,b". The output of the reducer, due the pass-through column, is sorted by "a". Additionally, it is sorted by "c" within each group (the last "a" in "c,a" is redundant) So the reducer would deliver sorting on "a,c"

public Collection<OrderingInfo> OrderingList { get; }

Public ReadOnly Property OrderingList As Collection(Of OrderingInfo)
	Get

public:
property Collection<OrderingInfo^>^ OrderingList {
	Collection<OrderingInfo^>^ get ();
}