On Wed, 19 Nov 2025 10:22:00 -0600, Paul Gilmartin <[email protected]> wrote:
>>What we've seen before is that people use varied locations for vendor >>products, > >Customers are arrogant, marking their territory like dogs. ROTFLOL, Please explain Linux sysplex and zVM sysplex before you call it arrogance. They don't exist and maint shoots 99.9999999 uptime to hell. Consider maint in Unix & zVM where everything is in a well named location (e.g. /usr/lpp/my_product) and it's applied to the live system. Nothing bad happens. Now consider maint on the live z/OS SYS1.LINKLIB. Not only do you risk crashing that system (e.g. LLA) but you risk crashing every system in that sysplex. We rename SYS1.LINKLIB by specifying volser (inactive) in SMP/e but there are times volser is impossible or impractical (e.g. /usr/lpp/my_product or SMS allocated datasets). Alternate names are critical to z/OS maint philosophy. >> so we're trying to be as flexible as possible while making it as simple as >> possible. >> >I suspect CALLLIBS is discouraged because SMP/E has no >way to know which LMODs need to be refreshed when a >CALLLIB is modified, probably outside SMP/E control. AUTOCALL only simplifies creating the PTF but it introduces risks. My personal opinion is that it should only be used for objects that are not included as part of the product (think sys1.miglib, language environment & ...). >I once noted here that when customers customize install >paths it invalidates instructions and examples in reference >materials. > >And Tech Support and consultants have the burden of >providing test cases to match the customized paths ROTFLOL, the alternative is for customers to give up on 99.9999999 uptime and sysplex. z/OS sysprogs deal all the time with customized dataset names and file locations. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
