Convert COMP-1 to Readable

gylbharat · Posted: Tue Aug 14, 2012 5:33 pm

Hi,

Is there a way to convert COMP-1 data type to readable format?

Like COMP-1 --> 1.2222E01

enrico-sorichetti · Posted: Tue Aug 14, 2012 5:35 pm

Nic Clouston · Posted: Tue Aug 14, 2012 5:36 pm

I presume you mean with SYNCSORT as JCL cannot do it and SYNCSORT questions come under this part of the forum.

Anuj Dhawan · Posted: Tue Aug 14, 2012 5:44 pm

gylbharat · Posted: Tue Aug 14, 2012 5:47 pm

I want to do it through Syncsort.

enrico-sorichetti · Posted: Tue Aug 14, 2012 5:50 pm

thanks &deity, couldn' t You have said so from the beginning

Bill O'Boyle · Posted: Tue Aug 14, 2012 6:18 pm

FWIW, this can be done easily in Assembler, COBOL and PL/I.

In fact, the same code can be used for COMP (Halfword/Fullword) and COMP-2 (Long Float).

Review the following - it's the same concept for all of the above data-formats, including COMP-1 -

www.ibmmainframes.com/viewtopic.php?p=172218&highlight=#172218

Anuj Dhawan · Posted: Tue Aug 14, 2012 6:34 pm

gylbharat- I wonder, you dont' remember this one: ibmmainframes.com/about55471.html

gylbharat · Posted: Thu Aug 16, 2012 12:10 pm

Hi,

I know, i raised this request some time back... but now its not FTP... we are doing a migration from VSAM files to DB2... we want to get the data in the readable format and then load data in DB2 through load utilities...

We want data to be in readable as it would also help in reconciliation.

Please let me know how we can convert COMP-1 and COMP-2 to readable format through SORT.

dbzTHEdinosauer · Posted: Thu Aug 16, 2012 1:11 pm

db2 has floating point (real and double - which is your comp-1 and comp-2 cobol types).

are you going to have db2 real and double or decimal datatypes?

my own question, but how are you going to reconcile?

maybe using the edit masks to convert floating point to 'readable'.
FL Floating point. Normalized. Signed.

gylbharat · Posted: Thu Aug 16, 2012 1:37 pm

Most probably we would be loading these values in Decimal data types..

Dick, Can you give me a sample to convert FL using EDIT MASKS to readable format?

dbzTHEdinosauer · Posted: Thu Aug 16, 2012 1:53 pm

sorry, i am not a syncsort user, i only have a manual.

gylbharat · Posted: Thu Aug 16, 2012 3:43 pm

Thanks Dick.... I will refer the manual... but a small sample would be helpful... may be someone who is syncsort expert can provide a sample

dick scherrer · Posted: Thu Aug 16, 2012 7:19 pm

Hello,

Syncsort MFX (1.4) supports Floating Point.

gylbharat · Posted: Thu Aug 23, 2012 4:01 pm

Any idea any one.... how do we convert comp-1 or comp-2 to readable format using sort?

dick scherrer · Posted: Thu Aug 23, 2012 7:16 pm

Hello,

If i understand, currently the floating point data is processed by the application. DB2 supports floating point (and the table(s) would surely store the data as floating point).

Why is there a need to create something "readable" in the middle? How does the application currently present these values so people can read them?

Bill Woodger · Posted: Thu Aug 23, 2012 8:22 pm

You're the one with Syncsort. Did you try with an edit mask? I agree with Dick and dbz, why bother making it "readable" (to humans with no knowledge of internal data representations). It sounds like DB2'll su** it up just fine as it is.

gylbharat · Posted: Fri Aug 24, 2012 5:17 pm

Hi,

Currently the floating point data is present in the VSAM file...

We will be writing a SORT to convert the VSAM file data to flat file with all readable data. That is expanding comp, comp-3, etc... Only problem is with the COMP-1 and COMP-2 fields where we are not able to convert it into readable format using sort.

Once we have the data in the readable format , we will FTP this data to Unix server, where it will be loaded in oracle table.

Having data in the readable format makes it easier in analysis and verify data.

Please let me know how we can convert the COMP-1 data to readable format?

Bill Woodger · Posted: Fri Aug 24, 2012 6:08 pm

OK, that is a very good reason to have it in character/text.

Have you tried the edit mask? Seems I've asked that before.

dick scherrer · Posted: Fri Aug 24, 2012 7:15 pm

Hello,

How does the current application present these fields to the user?

You may neeed to do the same thing here to get what you want. Possibly by coding a SORT Exit.

gylbharat · Posted: Mon Aug 27, 2012 1:29 pm

Hi,

I tried running with EDIT masks... But every data can be different depending upon power of 10. In such cases how do we specify how many digits there will be after decimal point.

Example.

INPUT FILE:

Bill Woodger · Posted: Mon Aug 27, 2012 1:37 pm

Easy. There will always be one digit in front of the decimal place. By convention, this will either be significati (1 through 9) or zero. The rest of (or the entire) the number is represented in the decimal part, and carries a "scaling" factor, like your example. E+09, E-02. Which is X10 to the power of 9, and X10 to the power of minus 2 (the last is a number less than 1 or greater than -1 if the number is negative.

If there is nothing in the MASK which supports this notation, then you can't do it with the MASK.

Which leaves, doing it in a program, or trying to "unpick" the data format. If you want to try the last, you have to find out how COMP-1 is stored (COMP-2 is the same, just longer).

gylbharat · Posted: Mon Aug 27, 2012 2:38 pm

Is there a way in sort to convert FL to External floating point as 1.23E+01

Bill Woodger · Posted: Mon Aug 27, 2012 2:52 pm

This is taking a long time :-)

You have Syncsort documentation. You can check whether there is anything in the edit MASK which supports floating point.

If there is not, you either have to "invent" a method to do it by decomposing the format of the data that the "float" is held in, or your need to do it in something else (as was suggested much earlier in the thread).

gylbharat · Posted: Mon Aug 27, 2012 3:31 pm

I do not have syncsort documentation... However, I checked the edit masks given in the below link...

www.ibmmainframes.com/post-40358.html

looks like there is no edit mask for floating point numbers.