| View previous topic :: View next topic |
| Author |
Message |
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
 Posted: Tue Apr 17, 2018 11:06 am Post subject: COBOL Picture Clause Editing S0C7 |
 |
|
Hello,
I have below code. When I provide input as "123!A" it doesnt give S0C7 but when I provide input as "12!3A" it gives S0C7. I understand I am providing incorrect input but just curious to know why in one case it gives 0C7. I was expecting it to fail in both cases.
| Code: |
DATA DIVISION.
WORKING-STORAGE SECTION.
01 WS-Z-VAR PIC ZZZZ9.
LINKAGE SECTION.
01 LS-PARM.
05 LS-PARM-LENGTH PIC S9(4) COMP.
05 LS-CHAR1 PIC X(05).
PROCEDURE DIVISION USING LS-PARM.
0000-MAIN.
DISPLAY 'LS-CHAR1 : >' LS-CHAR1 '< '
MOVE LS-CHAR1 TO WS-Z-VAR
DISPLAY 'WS-Z-VAR : >' WS-Z-VAR '< '
GOBACK. |
|
|
| Back to top |
|
 |
kolusu
Site Admin
Joined: 26 Nov 2002
Posts: 12370
Topics: 75
Location: San Jose
|
| Posted: Tue Apr 17, 2018 12:34 pm Post subject: |
|
|
rsantosh,
You need to understand how the compiler generates the instructions. Since you are moving a zoned decimal value to a zero suppressed output, the COBOL compiler will generate a PACK and ED instructions.
12!3A = X'F1F25AF3C1' will be packed in to 3 Byte value which will be of this format
X'A3' is an invalid packed decimal number and hence the S0C7
Now you passed
123!A = X'F1F2F35AC1' will be packed in to 3 Byte value which will be of this format
And X'3A' is a valid packed decimal number and hence no abend.
If you passed 1!23A = X'F15AF2F3C1' will be packed in to 3 Byte value which will be of this format
Notice that all are valid packed decimal values and hence you will not get an abend.
If you want to see the listing that the COBOL compiler generates then add the following at the top of cobol program( before the identification division) and see the sysprint listing.
_________________
Kolusu - DFSORT Development Team (IBM)
DFSORT is on the Web at:
www.ibm.com/storage/dfsort
www.linkedin.com/in/kolusu |
|
| Back to top |
|
|
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
| Posted: Tue Apr 17, 2018 11:36 pm Post subject: |
|
|
Hello Kolusu,
Thank you. Is a valid packed decimal number?
You shook my basic understanding of the packed decimal. 
All these years I used to think only the rightmost nibble(signed nibble) of packed decimal can contain A-F. Other nibbles must have 0-9 to make it a valid packed number.
So if a packed decimal has and I do arithmatic operation on it I wont receive S0C7?(will test this when I go to work later)
Santosh |
|
| Back to top |
|
|
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
| Posted: Tue Apr 17, 2018 11:42 pm Post subject: |
|
|
Hello Kolusu,
Or is it that EDIT tests for the validity of packed number byte wise? So on a whole 123A1C is an invalid packed decimal numebr but EDIT goes byte by byte and hence treats 3A (2nd byte) as a valid packed number?
Santosh |
|
| Back to top |
|
|
Nic Clouston
Advanced
Joined: 01 Feb 2007
Posts: 1075
Topics: 7
Location: At Home
|
| Posted: Wed Apr 18, 2018 6:06 am Post subject: |
|
|
what you have in this example is your original 'number' when packed is now two numbers:x'123A' and x'1C'. Both are valid but not as 1 number.
_________________
Utility and Program control cards are NOT, repeat NOT, JCL. |
|
| Back to top |
|
|
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
|
| Back to top |
|
|
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
| Posted: Wed Apr 18, 2018 6:45 am Post subject: |
|
|
| EDIT --> Assembler instruction. I compiled the program with LIST option and checked the generated instruction. |
|
| Back to top |
|
|
kolusu
Site Admin
Joined: 26 Nov 2002
Posts: 12370
Topics: 75
Location: San Jose
|
| Posted: Wed Apr 18, 2018 11:20 am Post subject: |
|
|
| rsantosh wrote: |
You shook my basic understanding of the packed decimal.
All these years I used to think only the rightmost nibble(signed nibble) of packed decimal can contain A-F. Other nibbles must have 0-9 to make it a valid packed number. |
rsantosh,
Your understanding is still correct, what I described is how the PACK instruction in assembler works.
To convert from zoned to packed
| Code: |
Format: label PACK D1(L1,B1),D2(L2,B2)
|
| PACK Instruction wrote: |
- Packs the L2 byte zoned decimal number at D2(B2)
and stores it as a L1 byte packed decimal number at D1(B1)
- The packing process:
1. The zone and numeric digit of the rightmost byte of
the zoned decimal number are reversed and placed in the
rightmost byte of the packed decimal number
2. The remaining numeric digits are moved to the packed decimal
number, proceeding from right to left. If the zoned
number has more digits than the packed number can hold, the
extras are ignored. If the zoned number has less digits
than the packed number, the remaining packed digits are
filled with zero
- The second operand does NOT have to be a valid zoned decimal
number
|
Suppose FLD1 is 4 packed bytes and FLD2 is 7 zoned bytes, with the
initial values:
| Code: |
FLD1 00 00 00 00
FLD2 F9 F8 F7 F6 F5 F4 F3
|
Execution of:
| Code: |
PACK FLD1(4),FLD2(7) will set FLD1 to 98 76 54 3F
PACK FLD1(3),FLD2(7) will set FLD1 to 76 54 3F 00
PACK FLD1(4),FLD2(5) will set FLD1 to 00 98 76 5F
|
| rsantosh wrote: |
and I do arithmatic operation on it I wont receive S0C7?(will test this when I go to work later)
|
Depends on how you are doing the arithmetic. You cannot perform arithmetic on numeric edited or character data. So you either need to redefine and do the arithmetic. So lets for a min assume that you did indeed redefined the value as zoned decimal, then now it comes down to how you are performing the arithmetic, whether you are doing arithmetic on each individual bytes or as a whole. It also depends on the compiler option NUMPROC whether the signs need to be considered.
| rsanthosh wrote: | | Or is it that EDIT tests for the validity of packed number byte wise? So on a whole 123A1C is an invalid packed decimal numebr but EDIT goes byte by byte and hence treats 3A (2nd byte) as a valid packed number? |
for the EDit instruction, the pattern and packed decimal number are both processed from left to right. The pattern is processed one BYTE at a time, while the packed decimal number is processed one DIGIT at a time.
A data exception (SOC 7) will occur if the second operand is not a valid packed decimal number.
_________________
Kolusu - DFSORT Development Team (IBM)
DFSORT is on the Web at:
www.ibm.com/storage/dfsort
www.linkedin.com/in/kolusu |
|
| Back to top |
|
|
rsantosh
Beginner
Joined: 20 Dec 2014
Posts: 38
Topics: 9
|
| Posted: Wed Apr 18, 2018 9:42 pm Post subject: |
|
|
Hello Kolusu,
Thank you for the detailed explanation. It is very useful.
Santosh |
|
| Back to top |
|
|
t-bonham@scc.net
Supermod
Joined: 18 Oct 2012
Posts: 30
Topics: 0
Location: Minneapolis, MN
|
| Posted: Thu Apr 19, 2018 8:28 pm Post subject: |
|
|
| rsantosh wrote: | Thank you for the detailed explanation. It is very useful.
Santosh |
Even more useful in a system, is tight editing of input data, so you don't end up with crap like this in a packed numeric field! |
|
| Back to top |
|
|
|
|
|
FAQ
Search
Quick Manuals
Register
Profile
Log in to check your private messages
Log in
