Static array query

noorkh · New User Joined: 06 Mar 2006 Posts: 76 Location: Chennai

Hi all,

I have used one static array in my cobol program.

Structure is as follows:

01 W-CLIENT-ARRAY.
03 W-CLIENT-COUNT PIC S9(4) COMP.
03 W-CLIENT-RECORD OCCURS 500 TIMES
INDEXED BY CLT-INDX.
05 W-CLIENT-CLT-NO PIC X(20).
05 W-CLIENT-KEY PIC S9(15) COMP-3.

Whenever i read record i will move to the above array. In the assumption that it will not be more than 500 records for the particular transaction i have declared it as occurs 500. But some of the scenario i got more than 500 records, in my last scenario i got around 506 records for a same transaction. I thought my program will abend as it exceeds 500 records but it went fine. And processed everything properly. Functionality also looks fine. Output files also created as expected.

Is that correct? Or it should abend?

Please help me.

Thanks,
Noor

William Thompson · Posted: Thu Dec 28, 2006 5:52 pm

Basically, you stepped on what ever storage lays beyond your array. There is a compile parameter that will check for that, bounds or some such I think, but you should be checking anyway.
When I write code and see some impossible condition I don't just ignore it because "It'll never happen".....I do the check and the imperative statement to the IF is abend with a abend code of "SNO#" where # is 0-9 and the SNO stands for "Should Not Occur"....

Bitneuker · Posted: Thu Dec 28, 2006 7:27 pm

So.......what William also tries to say is one shouldn't just be satified if the program seems to function correctly: wait till you sit at your desk in the middle of the night swetting to solve a dump/problem caused by some other developer's program who's done just the same

dcshnier · Posted: Thu Dec 28, 2006 10:49 pm

The advice shared by William is very good advice. Whenever you have created a fixed length array (i.e. 500 occurences in your example), always provide some code that will at least alert you to those rare situations when more than the stated occurences are encountered. As to why IBM did not automatically force an ABEND with a 'S0C4', we would have to know more detail. It seems to me that if you tried to move data into the 501st occurence of an array that only allowed for 500 occurences, you should have had a S0C4 ABEND.

A fringe benefit of taking William's advice is that the logic that you will add to check that if the array limit has exceeded, can be inserted before you attempt to move something into the array - that way you can pre-emptively ward off the S0C4 ABEND from happening to begin with. Furthermore, this logic does not have to result in a 'user-forced' ABEND. Instead, you can just issue a warning - or generate an email, or something less harsh than an ABEND. It all depends on how mission critical, will be the result of having the array being too small for that particular iteration.

regards
dcshnier

I

William Thompson · Posted: Thu Dec 28, 2006 11:12 pm

Bitneuker · Posted: Fri Dec 29, 2006 12:30 am

For a final 2 cents: Going beyond the indexed area doesn't defenitely mean a S0C4 since you might be poking in some working storage. S0C4 occurs when you are trying to ruin a protected area. A protected area usually contains pointers and machinecode generated from compilers. So the lesson to be learned is that, like both advisors wrote, defend you fucking boundaries before putting someone else into nighttrouble.

DavidatK · Posted: Fri Dec 29, 2006 3:52 am

Mr Bitneuker,

Do you truly need to use obscenities to express yourself? I find this site to be normally curious and friendly. Language as you have used degrades the site and certainly degrades yourself. By your post, I believe your English proficiency should be high enough to know better. There is no excuse to use this kind of language

William Thompson · Posted: Fri Dec 29, 2006 3:59 am

Oh Dave, relax, I read it as a somewhat over the top act, yes, sometimes you (as a knowledgeable system professional) you do have to "jump down their throats) to give them a little lesson (been there, done that, received the same).

DavidatK · Posted: Fri Dec 29, 2006 5:42 am

William,

I can remember a time, long ago, when IT was a 99% men, foul language and actions were common. But we have come far since then, women are in the work place now, thank goodness, and this is no longer acceptable. I personally find foul language in the work place (including user groups, and forums like this, etc.) unprofessional, either in the spoken word or the written word. Maybe up in hunting camp, or in a factory or on BLOGs, but not here. I find it in myself to overlook an occasionally incident that occurs in the heat of argument in the spoken word; the mouth move faster than the brain thinks. But for most of us, this is not the case with the written word.

I have ?never? used foul language to "jump down their throats) to give them a little lesson?, at least not in the last 25 years since I?ve matured some. It?s unprofessional, and almost always does not accomplish the goal. Don?t get me wrong, in the heat of an argument words have come out of my mouth that would make a teamster blush. But I cannot ever remember, except as a quote, using the ?F? word in correspondence.

True professionals, once shown the error of their way, feel bad enough without this kind of attack.

Just my philosophy, others have theirs.

DavidatK · Posted: Fri Dec 29, 2006 5:57 am

Noor,

As William pointed out, checking subscripts in arrays before populating is very important.

The following is a bit of code I put together that represents closely how I normally do table handling. I like to use subscripts for table handling instead of index by. Don?t ask me why, its just one of those life mysteries.

If you have and questions on the concepts, please let us know.

William Thompson · Posted: Fri Dec 29, 2006 7:03 am

DavidatK · Posted: Fri Dec 29, 2006 10:07 pm

William Thompson · Posted: Sat Dec 30, 2006 12:23 am

You missed the point, the reason indexes were created was to improve efficiency in certain table handling functions. To ignore using them where they shine best is to waste cpu time and you know what they say; "Time is Money".

Mickeydusaor · Posted: Sat Dec 30, 2006 3:58 am

Here is my 2 cents worth.

DavidatK · Posted: Sat Dec 30, 2006 4:10 am

Mickeydusaor,

Always need two more cents

I understand what you?re getting at; make the array so large that it could not possibly overflow. I've used this on single run 'FIX' programs because you don't have to program for all of these exceptions.

But, what happens when the program is in production for 10 years and the company growth, and acquisitions now puts the client base at more than 50000?

Have a happy holliday,

William Thompson · Posted: Sat Dec 30, 2006 4:14 am

Mickeydusaor,
I'm not quite sure what you are saying, but if that is all the logic and table handling, it seems a toss-up to me......

What happens to w-client-record after this? Is there a further redefinition of the table row?

Awaiting your next "2 cents"....

Bitneuker · Posted: Tue Jan 02, 2007 3:39 pm

Ok, considering Dave's remark a little while I feel I have to apologise using the 'F'-word; I'm not used to express myself like. What I actually tried is to tell what my thoughts were at the moment I finally found out why an application abended in the middle of the night. Some of you will certainly have similar experieces: get a call at 23.00, jump in your car to drive to the office, find out the abending program was written by a hired consultant who already left the company. At 03.00 we pinpointed the cause and solved it. The shiftleader then mentioned a similar problem occurred before; same consultant, same error. This consultant was at that time addressed to in order to point out what he caused and why, so he should have learned his lesson. After solving the recent problem we examined all of his programs and all of them still had the same error; though he was told the consequences of his 'bad' programming it seemed he didn't learn the lesson completely.

The remark was not addressed to TS (Topic Starter) so Noor, if I affended you please accept my apology.