FWIW (perhaps nothing), IBM solved the upgrade problem ~30 years ago for VM by separating the operating system nucleus (kernel) from the filesystem. That is, you can have multiple copies of the VM nucleus on various CMS (the end-user environment) minidisks, which CP (the hypervisor) knows how to read. To upgrade, you build a new VM nucleus on a new minidisk, then tell the standalone loader to IPL from that. If it fails, you just swap it back and go figure out the problem. It's super-slick. Not realistic for z/OS, I suspect, though I don't know enough to know why. And it was done as sort of a skunkworks project by David Boloker and Rich Corak in what was left of the Cambridge Scientific Center, in offices above the Copley Place mall IIRC. I doubt anyone could get away with that today; even at the time, I was surprised it was accepted into the base.
Windows has gotten better about upgrading, although, like most of you, I rarely upgrade a machine-usually by the time I'm forced to consider a new version, it's time for new hardware. But Windows upgrade difficulties mostly reflect the fact that the data isn't well separated from the OS: I still have a folder on my current machine called "Fromxxxx", where xxxx is the name of a machine four machines ago. This is itself a Bad Thing. Microsoft has tried, but then there's all the **** under c:\users\phsiii\Documents\ that isn't MY data, and cannot just be copied to a new machine. They should have separated "user data" and "installed stuff data" better. (I suspect the Registry was supposed to be the end-all here, but of course isn't.) It's fine for IBM to push toward a standardized layout. I don't think that's a bad thing at all, in principle-except for the existing shops who have zero time/resources/interest in "fixing" their configuration. Could all the PARMLIB stuff be pushed into one standard layout? Probably. Would it be easy? No, and it would make a lot of shops very nervous, I'm sure-"Yeah, Bob is the one who set that up before his heart attack, and nobody really knows the dependencies. I'm not touching it!" This isn't ideal, obviously, but it means that when an incremental change is needed, it can be made and tested in isolation, vs. some sort of big-bang reorg that's high-risk. It still sounds to me like the real problem here is that there hasn't been enough thought put into the impact on existing customers from the changes to install/upgrade beyond "Use z/OSMF". P.S. "Just watch this video"-sorry, that isn't how I (or, I suspect, many of the old-timers on here) do things. I don't have time to watch some video: I want a manual that explains the thing, so I can flip back and forth, add Post-Its, and/or copy/print excerpts. This is a growing trend that ignores that videos are a SLOW and inefficient way to learn many things; I attribute their rise to the fact that most people can't write a coherent sentence to save their lives, but they sure can talk. ...phsiii (sounding grumpy on a virtual Monday) ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
