Syncsort - COMP-2 to ZD

gylbharat · Posted: Wed Jul 27, 2011 6:24 pm

Hi,

I want to know how can we convert COMP-2 to ZD (Readable format) in Syncsort?

Suppose the value is 1.19875E3 it should get converted to 1198.75

gylbharat · Posted: Wed Jul 27, 2011 6:28 pm

I have referred to the post - ibmmainframes.com/about46065.html

But I don't want to loose the decimal part.

Bill Woodger · Posted: Wed Jul 27, 2011 6:40 pm

gylbharat · Posted: Thu Jul 28, 2011 1:31 pm

Its not working for me....

My value is

Bill Woodger · Posted: Thu Jul 28, 2011 4:35 pm

You know the drill. "It's not working" is useless.

You originally asked for it to go to ZD. ZD does not contain ".". You'd need an edit mask to get it to appear with the ".".

It seems you want six decimal places.

Your typical ZD in a sort has no implied decimal places (or, another way, it has any number of implied decimal places, you just have to do the implying seperate to the field).

So, say you have 10 bytes. You want six for the decimal places.

If you transfer the floating to the ZD, you will loose low-order significance unless you specifically code for it, as there is no implicit decimal in the ZD.

So, for six places, multiply by 1000000 first, then transfer to the ZD.

If this does not work, show what it does do, please.

gylbharat · Posted: Thu Jul 28, 2011 7:18 pm

Hi Bill,

I tried using this -

Bill Woodger · Posted: Thu Jul 28, 2011 7:46 pm

Mmmm.... I was thinking you wanted it in a ZD first, then could apply the edit the to ZD.

However, the approach you have taken might be hinting that this won't work either. On the other hand, you are getting more than one digit of integer.

Try just converting it to ZD so you know whether that route works. If not, back to the drawing board.

Have you got a SYNCSORT manual?

Bill Woodger · Posted: Fri Jul 29, 2011 1:03 am

gylbharat,

After I got a big hint, this is not going to work. Any arithmetic is only going to apply to the integer part of the floating-point number, so the fractional part cannot be made into "integer" by multiplication. Sorry about that.

Fancy a little Cobol to do it?

dick scherrer · Posted: Fri Jul 29, 2011 1:05 am

Hello,

Why might this need to be people-readable?

The computers have no trouble at all with the float format and if any real processing is to be done, i expect it is to be with computers.

gylbharat · Posted: Fri Jul 29, 2011 11:48 am

Hi Dick,

There is requirement where we have to FTP this file to a Unix box and there while processing this , the columns should be in a readable format.

dick scherrer · Posted: Fri Jul 29, 2011 7:38 pm

Hello,

Last i heard, UNIX works with floating point. . .

Probably there is something i misunderstand.

Bill Woodger · Posted: Fri Jul 29, 2011 8:05 pm

Hello,

Good old EBCDIC-to-ASCII translation. Good way to garble "non-display" fields, as all sorts of values for "control codes" in your binary fields get translated rather than left asis, let alone if you happen to have values which are "display" in your fields, they'll go mess about with themselves as well.

I know you're aware of this :-)

Also, is the Unix 7-bit? Is the float the same internal format? Best way to go for display, my opinion.