MVSFORUMS.com

[mcollins]

I have a file of records, with a key field and a value
There should be 2 records per 'transaction' ie the debit and the credit value
I want to ouput those records which have either
- a missing debit or credit,
- or where the values are not equal and complementary
(I was doing ok until it multiple pairings for the same key came in)
Can you suggest a method
e.g.
KEY1 5000+
KEY1 5000-
KEY1 5000+
KEY1 5000-
KEY1 1500+ --> output
KEY1 5000- --> output
KEY1 5000- --> output
KEY1 5000+
KEY1 5000-
KEY1 1200- --> output
KEY2 3000+
KEY2 3000-
KEY3 4000+ --> output
KEY4 4000+
KEY4 1200+ --> output
KEY4 4000-
KEY5 4000+ --> output
KEY5 3800- --> output

[kolusu]

mcollins,

The pairing is not consistent. take example of key 1

The first 4 records have a positivie and negative values which cancel out each other. But the next 6 records vary.

[mcollins]

Yep thats the problem the recs aren't consistent - I don't know how to best sort and then match them
from rec 5
rec5: 1500+ has nomatching 1500- so output
rec6: 5000- has no matching 5000+ so output
rec8 matches with either rec 7 or rec 9, so output the unmatched
rec10 has no matching +1200


The problem is the order of the file is random so ill have to sort them into some order
So taking your example and then making some of the pairings missing:
KEYA 5000+
KEYA 5000-
KEYA 4000+ unmatched
KEYA 3999+
KEYA 3999-
KEYA 3000- unmatched
KEYA 2000+
KEYA 2000-
KEYA 2000- unmatched
KEYA 1000- unmatched
KEYA 999+ unmatched

required Output
KEYA 4000+
KEYA 3000-
KEYA 2000-
KEYA 1000-
KEYA 999+


lrecl 100, recfm=FB
Key is 1,50
amt is posn 70 length 16 in format 000000000000.00s where s is + or -

[kolusu]

mcollins,

The following DFSORT/ICETOOL JCL will give you the desired results

[mcollins]

Wow... !
Thats amazing - thank you very much.
Am now just going step by step to see how it does it.
am loving the UFF/SFF