The rows in a table with the FIX attribute are fixed in length. Thus, when a table has many columns, processing efficiency is improved if the table is assigned the FIX attribute. In a sense, the entire row is manipulated as a single column. This is called manipulation on a row basis.

Manipulation of a table on a row basis provides the following advantages over manipulation on a column basis:

• Processing time is shorter.
• Processing time is unaffected by increases in the number of columns to be processed.
• Because one row can be transferred as one item of data, UAPs are easy to create and maintain.

If a table is assigned the FIX attribute, manipulating the table on a row basis rather than a column basis improves the processing efficiency for the following operations:

• When all or most of the columns will be retrieved
• When all or most of the columns will be updated
• When data will be inserted

Because an entire row is the target of manipulation, the embedded variables for transferring data must be re-declared if a column is added to the table.

A table with the FIX attribute cannot contain variable-length columns or null values. Therefore, manipulation on a row basis can be executed only for a table with no variable-length columns or null values. If manipulating an entire row as a single column will improve efficiency (particularly when there are many columns), consider eliminating columns with variable-length data and columns with null values; make the table attribute FIX. Use the following methods to do this:

• Look for short variable-length columns or variable-length columns where only a restricted portion stores data: convert them to fixed-length columns.
• Replace null values with some other values (e.g., 0 for numeric data and blanks for character data).