On Fri, 13 Oct 2017 14:18:29 -0500, Kirk Wolf wrote: >Pretty much any program can use z/OS Unix services. Strictly speaking I >don't think that there is an "inside" and an "outside", just "use". >So, just EXEC PGM=XXXXX and when the program makes its first Unix kernel >call it will be "dubbed" and get a Unix process id. That's about as much >as there is to establishing a program as a Unix process on z/OS. > >The tricky part comes why you try to redirect stdin/stdout/stderr from >processes to z/OS data sets, while providing for full dataset support. >That's where BPXBATCH sucks. In z/OS 1.7 BPXBATCH fixed stdout and stderr >redirection, but stdin is still broken. It isn't that hard if you know >how to use pipe() and select(). > BPXWUNIX provides full dataset support via DDNAMEs for stdin, stdout, and stderr. AOPBATCH (part of Infoprint) does likewise.
The fateful decision was made decades ago when IBM decided that MVS could not share ENQs between address spaces. To do so would have been a coding challenge then; vastly more difficult to retrofit nowadays. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
