|
View previous topic :: View next topic
|
| Author |
Message |
DB2 Guy
New User
Joined: 28 Oct 2008
Posts: 98
Location: Cubicle
|
|
|
|
Hi,
Lately there was an abend in prod due the "Division by zero" was raised. There is a calculation involved of below type, which caused the error:
| Code: |
COMPUTE something =
something1 / the-cluprit
END-COMPUTE |
One solution is to check for the value of "the-cluprit" and when it's >0 allow to execute the COMPUTE, however, business says that having zeros in this field is a valid business-condition, so if I use "zero-check" that actually violates this business condition.
I looked at the TRUNC(??) option of COBOL and it looks to me that if we use TRUNC(STD) instead of TRUNC(OPT), my problem can be solved. Prod code is compiled with TRUNC(OPT). I've tried to replicate this issue in test and working on it. Though manuals also talk about ON SIZE ERROR too.
I've also checked couple of base-line program which are uisng COMPUTE, some of them have ON SIZE ERROR coded and some not, though I could not check for the compiler option for all the programs...and thinking what actually can solve my problem with minimal changes involved.
It'd be nice if some one suggest, if I'm on right track and what others have used for similar circumstances.
Thanks. |
|
| Back to top |
|
 |
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10888
Location: italy
|
|
|
|
as usual the question is aimed at irrelevant technicalities...
| Quote: |
COMPUTE something = something1 / the-cluprit
END-COMPUTE |
| Quote: |
| however, business says that having zeros in this field is a valid business-condition, so if I use "zero-check" that actually violates this business condition. |
You are not violating anything
if the business condition allows a zero value somebody must tell what process to carry on
pseudocode ...
| Quote: |
if the-culprtit = 0 then
something = something??
else
something = something1/the-culprit
|
if the application had been properly designed/documented You would not have had the need to ask about zero division 
| Quote: |
| It'd be nice if some one suggest, if I'm on right track and what others have used for similar circumstances. |
sorry to disillusion but You are taking a long an windy road
the best approach is to fix the buggy application description
( You are just a victim of somebody' s else incompetence ) |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8700
Location: Dubuque, Iowa, USA
|
|
|
|
| Why would you think changing a compile option will suddenly resolve a divide by zero problem? There is absolutely no way that any compile option will resolve a divide by zero issue -- only application program changes will fix the issue, as enrico pointed out. |
|
| Back to top |
|
|
DB2 Guy
New User
Joined: 28 Oct 2008
Posts: 98
Location: Cubicle
|
|
|
|
Thanks enrico.
Robert,
Thanks for the reply.
I thought about the compiler option as the manual says:
| Quote: |
TRUNC(STD)
Use TRUNC(STD) to control the way arithmetic fields are truncated during MOVE and arithmetic operations. TRUNC(STD) applies only to USAGE BINARY receiving fields in MOVE statements and arithmetic expressions. When TRUNC(STD) is in effect, the final result of an arithmetic expression, or the sending field in the MOVE statement, is truncated to the number of digits in the PICTURE clause of the BINARY receiving field. |
Now, when a "divison by zero" occured, receving filed was not big enough to capture the result (infinity) and an on-size-error must have raised causing the abend. If, per the bold text in the quote, I use TRUNC(STD) then such an error should not happen? 
Regards, |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10888
Location: italy
|
|
|
|
what is that You did not understand in Robert and my advice!
You are on the wrong track
no cobol mangling will fix a badly defined business specification! |
|
| Back to top |
|
|
Bill O'Boyle
CICS Moderator
Joined: 14 Jan 2008
Posts: 2501
Location: Atlanta, Georgia, USA
|
|
|
|
A"Divide by Zero" exception "0C9" is raised when this occurs, regardless whether it's a binary or packed "Divide" instruction.
The TRUNC option only applies to binary data, but has nothing to do with testing or fixing a 0C9.
Even if your compiler supports COMP-5 (Native Binary --- TRUNC is ignored), it will still be raised when the divisor is ZERO, regardless.
Time to debug this beast and find the culprit divisor.
As far as violating the "business condition" condition, where ZERO is an acceptable value, you can do as you've suggested, issue the divide only when the divisor is non-zero. How would the business weasels know the difference? Just baffle them with technical jargon (a/k/a BS) and they'll go away with their tail between their legs.
That's about all you can do....
Bill |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8700
Location: Dubuque, Iowa, USA
|
|
|
|
| Divide by zero is not a valid mathematical operation. TRUNC cannot be used to fix this problem as there will be no "final result of an arithmetic expression" (to quote from your reference in the manual) -- the arithmetic computation fails before a final result can be calculated. |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10888
Location: italy
|
|
|
|
| seems that the topic heading nowhere, almost time to lock it! |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8700
Location: Dubuque, Iowa, USA
|
|
|
|
| Quote: |
seems that the topic heading nowhere, almost time to lock it!
|
enrico -- of course it's heading somewhere; can't you hear the flushing sound? |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10888
Location: italy
|
|
|
|
| my bad... I did not hear the sound of the chain being pulled |
|
| Back to top |
|
|
DB2 Guy
New User
Joined: 28 Oct 2008
Posts: 98
Location: Cubicle
|
|
|
|
Thanks and this thread does go in some direction, I've my answer.This
| Robert Sample wrote: |
| the arithmetic computation fails before a final result can be calculated. |
and
| Bill O'Boyle wrote: |
| As far as violating the "business condition" condition, where ZERO is an acceptable value, you can do as you've suggested, issue the divide only when the divisor is non-zero. How would the business weasels know the difference? Just baffle them with technical jargon (a/k/a BS) and they'll go away with their tail between their legs. |
Helps me to reach my answer.
Thanks for your patience. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19243
Location: Inside the Matrix
|
|
|
|
Hello,
Well, i'm only a little bit late to the party. . . 
An important missing bit of information is what process must be implemented when the-culprit is zero? Given that the divide by zero cannot happen, there needs to be a business rule that determines what to do when the-culprit is zero. . .
Good luck |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10888
Location: italy
|
|
|
|
| Quote: |
| Well, i'm only a little bit late to the party. . . |
You did not miss anything
as quite often happens on these forums the TS tweaked the replies in order to fit his expectations  |
|
| Back to top |
|
|
|
|
