On Thu, Sep 11, 2014 at 9:25 AM, Charles Mills <[email protected]> wrote: > Thanks all. > > 1. Source language is C++. I know how to call an "assembler" function from > C++. It seems odd to me that IBM would make such a basic OMVS function > available to assembler programs but only "by the way" available to C > programs, but whatever. > > 2. Definitely no Rexx involved. > > 3. I have this nagging fear that the problem is that because the program is > compiled #pragma RUNOPTS(POSIX(ON)) that as a result LE is failing it before > my first "user" instruction executes, and therefore I have no ability to > issue a more user-friendly message than a U4093 ABEND. I am trying to confirm > that. I am currently wrestling with an internal "you can't get there from > here" situation relative to testing with a no-OMVS-segment userid. Does > anyone happen to *know* whether this is the case? That POSIX(ON) means LE > requires an OMVS segment during program LE initialization? >
I cannot find a simple document which says something like: "A POSIX(ON) application requires that an OMVS segment exist for the RACF id which is running the application". I can see this: http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/ceea91c0/1.407 <quote> CEE3632I POSIX(ON) run-time option specified and the UNIX System Services feature is not available on the underlying operating system. </quote> http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/ceea91c0/1.531 <quote> CEE5002S POSIX function was not available. UNIX System Services were not started. </quote> >From which, one can conclude that using POSIX(ON) requires z/OS UNIX System Services be available. Which requires, at least in z/OS V1R13+, that the RACF user under which the program is running must have an OMVS segment. > I wish IBM wouldn't do this. Logically, a U4093/90 ABEND is no different from > an error message, but it does not work that way with customer psychology. We > send out a trial package, the customer runs the product (and yes, we document > the need for an OMS segment, but who reads documentation? Not our customers.) > and it gets a U4093 ABEND. Does the customer look it up? No, he picks up the > phone, calls the salesman and says "your product blows up when we try to run > it." The salesman tells support. By then the customer has deleted the SYSOUT > without noting the exact ABEND, so we have to persuade the disgruntled > prospect to run the product AGAIN -- he inevitably reports "no better -- it's > STILL blowing up" so we can get the ABEND code and diagnose the problem. In > fact, get all of the symptoms because at that point we don't know that the > ABEND number will be sufficient to diagnose the problem. A nice, readable > message, either from LE or from us, would be a lot better all around. End of > rant. > I agree that a message to the effect of "RUNNING A POSIX(ON) APPLICATION REQUIRES AN OMVS SEGMENT" would be nice. I guess that today's businesses simply don't want to understand anything outside of their specific market. That is why our company executives have said "We are in the health business, not the IT business". Which is what is driving them to using ?aaS (SaaS, IaaS, PaaS, etc) in preference to having on-site hardware/software beyond the desktop. In the old days, this was lumped under "time sharing" and "outsourcing". I am guessing, and that's all it is, that eventually this company will have only application programmers and _maybe_ a group to do telephony and desktop support. They might even try to use "as needed" consultants for that. (like Geek'R'Us or something). > Charles -- There is nothing more pleasant than traveling and meeting new people! Genghis Khan Maranatha! <>< John McKown ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
