With row partitioning (for a PPI or column-partitioned table), the Teradata Database makes sure rows are placed in their appropriate partitions.  When the row partitioning for the table is altered, rows may need to move from one partition to another so they are in their appropriate partitions according to the altered partitioning.  The partitioning expression must be deterministic (always computes the same value for the same value of the partitioning column) to properly place and retrieve the row.

For instance if the table has RANGE_N partitioning, ALTER TABLE will move rows automatically from NO RANGE to newly-added ranges as needed.  Also, if a range is dropped, rows from that partition are moved to the NO RANGE, NO RANGE OR UNKNOWN partition or a newly-added range that covers the dropped range.