In a recent note, Farley, Peter x23353 said:
> Date: Thu, 1 Dec 2005 17:32:39 -0500
>
> I had thought many years ago that IBM was on a path to make interpreted REXX
> the replacement language for controlling all jobs, but I was sadly mistaken
> and VM-centric in that thought -- I didn't count on the recalcitrance of the
> MVS-centric hierarchy at IBM.
>
But that hierarchy has embraced Rexx, even if at shotgun-point.
What's needed to make Rexx an alternative (I won't frighten
readers by saying "replacement") for controlling some jobs?
o Unrestricted ability to run APF-authorized programs; possibly
by launching them in a separate address space, as BPX1EXM
does, but without the allocation restrictions of BPX1EXM.
o A facility for performing SYSDSN ENQs en masse, in advance
of individual allocation.
- such a facility must not require APF authorization.
- the facility must wait for ENQ resolution, again without
requiring APF authorization.
- the facility must detect and reject any ENQ request that
creates a deadlock. I'll accept that "deadlock" means
any creation of a cycle in the graph of ENQ dependencies,
regardless that asynchronous activity in one of the nodes
might potentially later resolve the deadlock.
o Facilities that I'm accustomed to for handling tapes,
particularly RETAIN and VOL=REF.
A Rexx-based alternative might take the form of a Zen-like
unasking of the FAQ: "How can I determine the return code of
a prior job step, or of a job step in a submitted job?"
Suppose a job, written in Classic JCL could be submitted to
run not autonomously, as when submitted via INTRDR, but
preserving a communication channel (socket, POSIX pipe,
batchpipe, whatever) to the submitter's address space.
Such a channel need transfer little information beyond
step name and completion code; it would serve largely to
synchronize the submitter, which could then get more
detailed information from data sets created by the job
step(s). To this end, temporary data sets should not be
deleted until the submitting address space severs the
connection.
-- gil
--
StorageTek
INFORMATION made POWERFUL
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
