Feed the dataset list into a REXX which generates ISPF edits with an edit macro to do the work. This could even run as an ISPF batch job.
I believe that DF/SORT can also do uppercase translation, if so you need to generate a job to run the SORTs. ISPF skeleton services could do this fairly easily and the skeleton builder could run as an ISPF batch job.
If I use Sort, I create a new dataset name and delete the old, correct?
Willy Jensen - a few postings above - has demonstrated very clearly: SORT can either create a new (copy) dataset, or override the old one. It's up to you.
ISPF EDIT usually makes the updates in-place; but with some tricks it also can create a copy of a dataset.
I have seen IBM advising not to use the same dataset for input and output, but I have done this lots of times over the years without any problems, so it might just be their lawyers speaking.
You can use the job here as a sample of generating the LTOU job using a batch ISPF skeleton builder. The dataset list is referenced by ddname DSLIST in the ISPBAT step. Of course, you could move the members in step 1 to SYSEXEC and ISPSLIB if this has to be done on a regular basis. It also demonstrates some of the lesser know features of skeletons, the ability to use TSO commands and the stack.
Joined: 15 Aug 2015 Posts: 1379 Location: Bamberg, Germany
We should be clear if the input is all of fixed length or variable. Depending on that, the method has to be chosen. Fixed length, I would chose a SORT PROC, otherwise EDIT MACRO, or whatever the USS hell has spawned to achieve the task (tr/sed/awk..).