No. Use the ANSI windowed format if this is what you intend:
RANK() OVER (PARTITION BY Account_Number, Date_Added, Time_Added)
In the (deprecated) Teradata RANK syntax, the columns only specify sort order; i.e. your query is effectively computing
RANK() OVER (ORDER BY Account_Number, Date_Added, Time_Added)
And regarding your question on MERGE, all columns in the PI/PPI must be directly specified with an equality predicate in the ON clause. So it is not possible to "match" a NULL value in one of those columns.
↧