Need to get all the records of all the occurances

GaniKrish143 · New User Joined: 30 Jan 2013 Posts: 4 Location: India

Hi Everybody,

I have an issue in my cobol program.

The program takes 2 input files and gives 2 ouput files.
one file which will have account numbers
other will have acccount numbers corresponding records.

to get the records wrt to the account number, I have given the table and search concept.

The problem is when am searching for a account number in input file am getting the last occurence of records correspondint to accountno, I need to get all the records of all the occurances of accountno.

This is the search logic I have used.

enrico-sorichetti · Posted: Wed Jan 30, 2013 2:58 pm

USE SORTED FILES
and look at the TWO files match program here
www.ibmmainframes.com/viewtopic.php?t=22649

GaniKrish143 · New User Joined: 30 Jan 2013 Posts: 4 Location: India

Hi Entico,

I did not asked for program for matching files.

my requirement is I have the input file which consists of records(80 in lengh) corresponding to each account.

in the second input file I have only the accounts(10 in length) which is coming from prevous step in jcl(the accounts which has the problems).

In the input file this account may occurs so many times, so I need all the records of all occurences of the accountno which is in second input file.

for this I loaded the first input file into a table and the used search concept in first input file using the accountno in the second input file, if a match occurs am moving the corresponding records to output file.

The problem is in search, while in linear search we will set the index to 1, to start from first position.
for example take this scenario, if a match encounters in 49th position, 65 th position, 74 th position and 98 th position.

when writing the output record it is moving only 98 th position records.
not all the matched records.

this is the issue.

and this is the code in search

2400-SEARCH-RACT.
SET REIINDX TO 1
SEARCH REI-TABLE
AT END
MOVE 'N' TO WS-RACT-FOUND-SW
WHEN
WS-REI-TBL-KEY (REIINDX) = WS-ACT-CHECK
MOVE 'Y' TO WS-RACT-FOUND-SW
END-SEARCH.
2400-EXIT.
EXIT.

Ranjithkumar · New User Joined: 10 Sep 2008 Posts: 93 Location: India

You can make the input file a VSAM (ESDS as you have multiple occurrences for each account). Then use START BROWSE / READ NEXT based on the account from the other file , to achieve what you want. This should be efficient than using the SEARCH.

Bill Woodger · Posted: Wed Jan 30, 2013 4:08 pm

enrico's suggestion is much better than what you are doing, whether you asked for it or not. Sorry about that.

Are you saying your File 1 is the largest of the two files you are processing, and may have multiple occurrences (as records) of account-number, distributed across the file? And File 2 is just a list of, unique, account numbers you wish to process?

And then you decided to store the large file with the non-unique occurrences and SEARCH that?

Bill Woodger · Posted: Wed Jan 30, 2013 4:09 pm

Ranjithkumar · New User Joined: 10 Sep 2008 Posts: 93 Location: India

Bill

I am not suggesting to make it a CICS program. I am suggesting to make the input file VSAM. By START BROWSE i do not mean the CICS STARTBR command. Its the same logic in batch, START the file using account number, read the next record until the key changes.

Bill Woodger · Posted: Wed Jan 30, 2013 4:48 pm

So why make BROWSE "look" the same as START? BROWSE means nothing in Cobol.

Now, as to your suggestion re-described. You want to make the dataset VSAM. An ESDS. You want to use START. So you need an ESDS with an alternate index.

And without any information on numbers of records or percentage of "hits" you'd just go with this? Mmmm....

chandan.inst · Posted: Wed Jan 30, 2013 4:51 pm

Hi,,

Have you tried using PERFORM- VARYING concept fro searching the table like below
But this will not be good from Performance point of view

enrico-sorichetti · Posted: Wed Jan 30, 2013 5:14 pm

GaniKrish143 · New User Joined: 30 Jan 2013 Posts: 4 Location: India

sorry Enrico, its type and am new to this site so bit tensed.
anyways Thanks for your valuable sugession

Robert Sample · Posted: Wed Jan 30, 2013 5:24 pm

GaniKrish143, based on what you have posted so far, your design is broken. A better design would be to sort your first file so the records for a given account are sequential, not scattered throughout the entire file.

If the accounts are scattered across the file, you do not want to use a SEARCH verb. Manually search the entire table looking for matches.

GaniKrish143 · New User Joined: 30 Jan 2013 Posts: 4 Location: India

I tried with Chandan code
its working fine for me

Thank you all of yoU

for giving replies

chandan.inst · Posted: Wed Jan 30, 2013 6:46 pm

Good to know that its working for you..

Regards,
Chandan