The DataTableReader works much like any other data reader, such as the SqlDataReader, except that the DataTableReader provides for iterating over rows in a DataTable. In other words, it provides for iterating over rows in a cache. The cached data can be modified while the DataTableReader is active, and the reader automatically maintains its position.
When you create a DataTableReader from a DataTable, the resulting DataTableReader object contains one result set with the same data as the DataTable from which it was created, except for any rows that have been marked as deleted. The columns appear in the same order as in the original DataTable. The structure of the returned result is identical in schema and data to the original DataTable. A DataTableReader that was created by calling the GetDataReader()()() method of a DataSet object contains multiple result sets if the DataSet contains more than one table. The results are in the same sequence as the DataTable objects in the DataTableCollection of the DataSet object.
The returned result set contains only the current version of each DataRow; rows that are marked for deletion are skipped.
The DataTableReader provides a stable iterator; that is, the contents of the DataTableReader are not invalidated if the size of the underlying collection is modified during iteration. For example, if one or more rows in the Rows collection are deleted or removed during iteration, the current position within the DataTableReader is maintained appropriately and it does not invalidate the iterator.