DISPLAY in COBOL of Signed Comp-3 Data shows unexpected output
Theoretically, I studied like the end of character in comp-3 tells whether it is positive or negative value
- C - Indicates positive value
- D - Indicates negative value
Is this not applicable for new version of COBOL in mainframes?
01 WS-COMP3 PIC S9(5) COMP-3 VALUES -12.
DISPLAY WS-COMP3
OUTPUT: 0001K
For above piece of code, I am getting the end of character as K instead of D The value K is the substitution of -2 0 ==> } -1 ==> J, -2 ==> K ....
Using DISPLAY ... with a numeric data type requires a conversion to a displayable type. The COBOL standard requires it.
A typical conversion for COMP-3 is to move the data item to an equivalent displayable format. For this case, PIC S9(5) COMP-3 is often converted to PIC S9(5) SIGN TRAILING for display.
This conversion means the internally stored value will be converted so that individual digits, except the last, will be converted to displayable digits. The last will have the sign indicator changed to reflect the format for the particular implementation.
For IBM mainframes, the internal COMP-3 format for -12 is 00 01 2D and will be converted to F0 F0 F0 F1 D2 which displays as 0001K.
Many ASCII systems will provide a slightly different result. The same internal format will be converted to 30 30 30 31 x2 where the x depends on the implementation's requirement. It may display as 0001B or 0001r or some other, such as SIGN SEPARATEgiving -00012.
The actual conversion for any data type done by any COBOL implementation will be documented in the language reference.
From the 2002 standard, B.1 Implementor-defined language element list,
- DISPLAY statement (data conversion). This item is required. This item shall be documented in the implementor's user documentation. (14.8.10, DISPLAY statement, general rule 1)
DISPLAY statement, 14.8.10.3 General rules,
- The DISPLAY statement causes the content of each operand to be transferred to the hardware device in the order listed. If an operand is a zero-length data item, no data is transferred for that operand. Any conversion of data required between literal-1 or the data item referenced by identifier-1 and the hardware device is defined by the implementor.
As an addition to Rick Smith's excellent answer describing the reasons I wanted to add that IBM's Enterprise COBOL for z/OS since version 5 provides a compiler-option to handle this issue.
When compiling with DISPSIGN(SEP) a DISPLAY of a signed numeric item (binary, packed decimal or zoned) will always produce a separate leading sign.
Default is DISPSIGN(COMPAT) which will behave like shown in the question.