Okay, I think I have it. I've spent too much time on this already.

Yeah, it's imperfect. 

- The problem Gil describes below
- TF seems confused by the binary data and at the very least stops on any line 
with a leading blank, I think.
- Maybe one or two other details.

But it's good enough for test data. I can just delete any hosed up records.

The trick is defining a PF key as :TS;RFIND;RFIND

That pretty well automates everything except the eyeball check "is this a real 
record?" You need two RFINDs because the first one finds the same byte again 
because it has now been moved down onto a lower line by TS. 

A little bit of a PITA: more records than I thought. Oh well.

Sorry for hitting the panic button about trailing blanks. I was not familiar 
with the look of the ISPF editor with V records and HEX ON. It seems to me that 
the trailing blanks it displays are wrong: there are not blanks there; there is 
nothing there, which is not the same thing. Whatever.

No idea why ISPF is blowing up with a S0C4. Very strange. The sequence is as 
follows:

Edit data set -- all goes as expected. Exit out of edit.
Edit data set -- ISPF immediately S0C4's.
Edit data set again -- all goes as expected.
Repeat as necessary.

Thanks all!

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf 
Of Paul Gilmartin
Sent: Tuesday, January 15, 2013 12:44 PM
To: IBM-MAIN@LISTSERV.UA.EDU
Subject: Re: Break a dataset into new record boundaries?

On Tue, 15 Jan 2013 12:16:55 -0800, Charles Mills wrote:

>Yeah, it seems to be working. Not sure yet. I think perhaps the apparent 
>blanks are just ISPF edit window dressing at the ends of the lines. It doesn't 
>know about short lines.
>
Of course, since you say the data are binary, a 0x40 at the end of a record (or 
where one was inadvertently broken) may be critical.  Unfortunately, there's no 
way to tell ISPF not to add or remove trailing blanks.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN

Reply via email to