On Sun, 7 Sep 2025 13:36:25 -0500, Jon Perryman wrote: > >Pipes 101: The most rudimentary z/OS PIPE is batch (step 1 feeds step2 that >feeds ... that feeds step #). > Can batch do this without using a temporary data set?
can a downstream step take timely action on output from an upstream step? This is routine design in CMS Pipelines. >Multi-tasking 102: TSO and ISPF do not multi-task because people do not >understand multi-tasking and would abuse it. TSO PIPEs are not acceptable >because it encourages mult-tasking. > I hear the arrogance alarm. Don't thus demean your colleagues. >CMS pipes 101: CMS PIPE's power is in the CMS pipe subcommands built >specifically for PIPE because CMS has very few commands and they were not >designed with PIPEs in mind. > Agreed. CMS facilities are meager. I/O abstraction is sorely lacking. Pipelines fills the gap. But Pipelines can not run two CMS stages concurrently. MVS ATTACH does that well, with some restrictions due to the kludge of the alternate DDNAME list. It should have been possible to make DDNAMEs have task scope optionally, instead of job scope. >TSO pipes 101: Those CMS PIPE subcommands could be implemented as TSO >commands. They could even be implemented as programs called using ADDRESS >LINK. If you can't easily solve the same problems as CMS, then you're asking >the wrong person for help. > ADDRESS ATTCHMVS for compatibility. But, ironically, that does not support multitasking, at which some people, contrary to your prejudice, are competent. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
