On 29 August 2013 16:18, R.S. <r.skoru...@bremultibank.com.pl> wrote: > I just found the following in some IBM same JCL (job, actually): > //SYSUT1 DD UNIT=(SYSDA,SEP=(SORTLIB,SYSLMOD,SYSLIN)), > // SPACE=(1000,(60,20)) > > Last change date is half of the 2013 (creation date is probably 2005 or so). > As far as I know SEP is syntax checked and ignored for many moons, at least > since first OS/390, but I vaguely remember somebody told me it was obsoleted > with advent of MVS.
I slightly more than vaguely remember it the same way. SEP= disappeared (was ignored) with MVS, i.e. OS/VS2 2.0, because it limited the then-new SRM's ability to swap a job in or out to control the I/O workload on a channel. There is an element of conflict between the application programmer/job owner's wish to have the best possible performance for an individual job by spreading its I/O around, vs the SRM's wish to control overall system performance as specified by the performance group definitions. Swapping out or in a job with allocations all over the place would not predictably change the load on an over or under busy channel. And on that note, I even more vaguely remember that SEP= (in pre-MVS OSs) was done at the channel level rather then the device, but that's so vague as to be unreliable. Since the source code for MVT and MVS 3.8 are both available online, anyone sufficiently interested can read the code and report back. On the matter of a recent piece of JCL containing these obsolete keywords, well that is all too common. There is a widespread and hard to break culture consisting of some blend of "if it ain't broke, don't fix it", "I'll just copy this JCL that works and change it minimally to suit my needs", and "we're not really sure we need this anymore, but I don't want to be the one to remove it and find it breaks something important, so I'd better leave it in". I removed SEP= from one of our product manuals and around 2000, and when we acquired another product almost ten years later, I found its install manual to have SEP= too. I may yet have missed some... Tony H. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
