Natural for Mainframes Version 8.2.5 for Mainframes
 —  Programming Guide  —

Arrays

Natural supports the processing of arrays. Arrays are multi-dimensional tables, that is, two or more logically related elements identified under a single name. Arrays can consist of single data elements of multiple dimensions or hierarchical data structures which contain repetitive structures or individual elements.

This document covers the following topics:


Defining Arrays

In Natural, an array can be one-, two- or three-dimensional. It can be an independent variable, part of a larger data structure or part of a database view.

Important:
Dynamic variables are not allowed in an array definition.

Start of instruction set To define a one-dimensional array

Start of instruction set To define a two-dimensional array

Top of page

Initial Values for Arrays

To assign initial values to one or more occurrences of an array, you use an INIT specification, similar to that for "ordinary" variables, as shown in the following examples.

Top of page

Assigning Initial Values to One-Dimensional Arrays

The following examples illustrate how initial values are assigned to a one-dimensional array.

Top of page

Assigning Initial Values to Two-Dimensional Arrays

This section illustrates how initial values are assigned to a two-dimensional array. The following topics are covered:

Preliminary Information

For the examples shown in this section, let us assume a two-dimensional array with three occurrences in the first dimension ("rows") and four occurrences in the second dimension ("columns"):

1 #ARRAY (A1/1:3,1:4)

Vertical: First Dimension (1:3), Horizontal: Second Dimension (1:4):

(1,1) (1,2) (1,3) (1,4)
(2,1) (2,2) (2,3) (2,4)
(3,1) (3,2) (3,3) (3,4)

The first set of examples illustrates how the same initial value is assigned to occurrences of a two-dimensional array; the second set of examples illustrates how different initial values are assigned.

In the examples, please note in particular the usage of the notations * and V. Both notations refer to all occurrences of the dimension concerned: * indicates that all occurrences in that dimension are initialized with the same value, while V indicates that all occurrences in that dimension are initialized with different values.

Assigning the Same Value

Assigning Different Values

Top of page

A Three-Dimensional Array

A three-dimensional array could be visualized as follows:

The array illustrated here would be defined as follows (at the same time assigning an initial value to the highlighted field in Row 1, Column 2, Plane 2):

DEFINE DATA LOCAL  
1 #ARRAY2  
  2 #ROW (1:4) 
    3 #COLUMN (1:3) 
      4 #PLANE (1:3)  
        5 #FIELD2 (P3) INIT (1,2,2) <100> 
END-DEFINE 
...

If defined as a local data area in the source editor, the same array would look as follows:

I T L Name                             F Leng Index/Init/EM/Name/Comment       
- - - -------------------------------- - ---- ---------------------------------
    1 #ARRAY2                                                                  
    2 #ROW                                    (1:4)                            
    3 #COLUMN                                 (1:3)                            
    4 #PLANE                                  (1:3)                            
I   5 #FIELD2                          P    3                    

Top of page

Arrays as Part of a Larger Data Structure

The multiple dimensions of an array make it possible to define data structures analogous to COBOL or PL1 structures.

Example:

DEFINE DATA LOCAL  
1 #AREA 
  2 #FIELD1 (A10) 
  2 #GROUP1 (1:10)  
    3 #FIELD2 (P2)  
    3 #FIELD3 (N1/1:4)  
END-DEFINE 
...

In this example, the data area #AREA has a total size of:

10 + (10 * (2 + (1 * 4))) bytes = 70 bytes

#FIELD1 is alphanumeric and 10 bytes long. #GROUP1 is the name of a sub-area within #AREA, which consists of 2 fields and has 10 occurrences. #FIELD2 is packed numeric, length 2. #FIELD3 is the second field of #GROUP1 with four occurrences, and is numeric, length 1.

To reference a particular occurrence of #FIELD3, two indices are required: first, the occurrence of #GROUP1 must be specified, and second, the particular occurrence of #FIELD3 must also be specified. For example, in an ADD statement later in the same program, #FIELD3 would be referenced as follows:

ADD 2 TO #FIELD3 (3,2)

Top of page

Database Arrays

Adabas supports array structures within the database in the form of multiple-value fields and periodic groups. These are described under Database Arrays.

The following example shows a DEFINE DATA view containing a multiple-value field:

DEFINE DATA LOCAL  
1 EMPLOYEES-VIEW VIEW OF EMPLOYEES  
  2 NAME 
  2 ADDRESS-LINE (1:10) /* <--- MULTIPLE-VALUE FIELD 
END-DEFINE 
...

The same view in a local data area would look as follows:

I T L Name                             F Leng Index/Init/EM/Name/Comment       
- - - -------------------------------- - ---- ---------------------------------
  V 1 EMPLOYEES-VIEW                          EMPLOYEES                        
    2 NAME                             A   20                                  
  M 2 ADDRESS-LINE                     A   20 (1:10) /* MU-FIELD

Top of page

Using Arithmetic Expressions in Index Notation

A simple arithmetic expression may also be used to express a range of occurrences in an array.

Examples:

MA (I:I+5) Values of the field MA are referenced, beginning with value I and ending with value I+5.
MA (I+2:J-3) Values of the field MA are referenced, beginning with value I+2 and ending with value J-3.

Only the arithmetic operators plus (+) and minus (-) may be used in index expressions.

Top of page

Arithmetic Support for Arrays

Arithmetic support for arrays include operations at array level, at row/column level, and at individual element level.

Only simple arithmetic expressions are permitted with array variables, with only one or two operands and an optional third variable as the receiving field.

Only the arithmetic operators plus (+) and minus (-) are allowed for expressions defining index ranges.

Examples of Array Arithmetics

The following examples assume the following field definitions:

DEFINE DATA LOCAL 
01 #A (N5/1:10,1:10) 
01 #B (N5/1:10,1:10) 
01 #C (N5) 
END-DEFINE 
...
  1. ADD #A(*,*) TO #B(*,*)

    The result operand, array #B, contains the addition, element by element, of the array #A and the original value of array #B.

  2. ADD 4 TO #A(*,2)

    The second column of the array #A is replaced by its original value plus 4.

  3. ADD 2 TO #A(2,*)

    The second row of the array #A is replaced by its original value plus 2.

  4. ADD #A(2,*) TO #B(4,*)

    The value of the second row of array #A is added to the fourth row of array #B.

  5. ADD #A(2,*) TO #B(*,2)

    This is an illegal operation and will result in a syntax error. Rows may only be added to rows and columns to columns.

  6. ADD #A(2,*) TO #C

    All values in the second row of the array #A are added to the scalar value #C.

  7. ADD #A(2,5:7) TO #C

    The fifth, sixth, and seventh column values of the second row of array #A are added to the scalar value #C.

Top of page