On 19/06/2022 1:33 am, Paul Gilmartin wrote:
On Sat, 18 Jun 2022 09:51:45 -0700, Charles Mills wrote:... I picture writing the started task in Rexx, so I would have to write to a DD name allocated to the UNIX file (either dynamically or with JCL), not with "native" C fopen(), fwrite(), etc. Does that change any of the answers?Why? In Rexx you can "address SYASCALL write ..." instead.
Why REXX? Is it a case of knowing the banjo so you play Stairway to Heaven in the style of Earl Schruggs?
Why not use IBMs z/OS Python? You can then use SQLite instead of a file which will significantly simplify writing reports. In fact, it would trivial to serve those reports as a REST API and put a nice WebUI on top using a simple template that supports data tables.
There seems to be a convention where the reporting process rename()s the log file. The logging process (daemon?) (task?!) continues to write to the open descriptor until it closes it, conventionally by a SIGHUP(!?) from the reporting process. The logger then open( ,O_CREAT + O_APPEND)s a new instance and the reporting process reports from the renamed one. No data drops through a timing window <https://en.wikipedia.org/wiki/SIGHUP#Modern_usage>. o When I wear my MVS hat, I can't grok fork(). o When I wear my UNIX hat, I can't grok ENQ. o If I put on both hats, my brain overheats.
---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
