By 'temporary' I mean short-lived, where a file is created and cataloged in a job stream and then deleted either at the end of the job or at the start of the next run. True temporary data sets in the MVS sense are not cataloged anyway, so they're not pertinent to this discussion. My point was about degree of catalog activity. USS files are typically created once and (almost) never deleted. And yes, it might be nice to have true temporary USS files.
As for the bewarish warnings, I'll pretend I never heard them. ;-) . . . J.O.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [email protected] -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Paul Gilmartin Sent: Thursday, August 20, 2015 9:52 AM To: [email protected] Subject: Re: Catalog for OMVS datasets On Thu, 20 Aug 2015 16:26:30 +0000, J O Skip Robinson wrote: >No unique catalog for USS objects. System level OMVS all have the same HLQ >pointing to a general purpose user catalog. Individual USS files typically >have the user's HLQ or some other HLQ associated with the application; these >all go into the user catalog associated with the alias. > But beware. There are warnings that using user HLQs for UNIX repositories engenders an opportunity for escalation of privileges, etc. Provided the user can reverse-engineer the HFS/zFS format, dismount the UNIX FS, and tweak the SUID bit, etc. (Do mount maps provide a NOSUID attribute?) >I've never investigated this, but I suspect that HFS/ZFS in general are much >more stable than traditional MVS files. No GDGs, no temporary allocate/delete >processes in batch. Allocate once, use forever at the MVS file level because >the dynamic activity is all in the USS directory structure. I see no reason >for a unique catalog. > Temporary data sets (are we discussing the same thing?) are among the Classic features I miss most in UNIX. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
