This is my main difficulty with CA-MSM. It is great for doing the download from CA and the SMP/E tp the point of install and tailoring. But, then, you need to do it CA's way, not our way. I concede that our way may not be the best way, but we have a lot invested in some parts of it. My other objection is that the tailor and deploy processes make copies of "targets" not actually required for the final execution of the product. Our way only copies the specific libraries and data needed to run.
As for zOSMF, today I only have kneecapped CP's and JAVA is not known to perform well. > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[email protected]] > On Behalf Of Paul Gilmartin > Sent: Tuesday, October 04, 2016 10:35 AM > To: [email protected] > Subject: Re: ISVs, SMP/E, ... (was: Installation Improvements ...) > > On Tue, 4 Oct 2016 09:15:33 -0700, Charles Mills wrote: > > > >- we have no one who is a drop-dead SMP/E packaging design expert. > >- when we go to people who are SMP/E packaging experts they say "your > IEBCOPY install is a work of art! It is so simple! Why would you want to > change? SMP/E is waaaaay overkill." > >- but some larger customers really, really, really want SMP/E. One Sysprog > said something to me like "yeah, SMP/E is awful, but it's the same awful for > everything. (Goes to Tom Brennan's point.) I'm sure your install is great, but > it's unique, and I have limited learning time." > > > One source of complexity is the habit of customers to insist on paths/DSNs > that follow local conventions. "Ya gotta have standards!" But the customer > and the ISV have different views of who is better positioned to define the > standards. > > I once said among these fora that if the customer chooses the names the > examples in the supplied documentation are invalidated; they can't be simply > copied-and-pasted. The rebuttal came, "The docs should be machine > readable and the customer can simply run them through a filter that updates > the names." Spare me! I hope he was facetious. > > IBM took a step in a good direction by providing the /usr/lpp hirearchy (I > would have preferred /usr/local.) Something similar should have been done > for Classic data set names long ago. (44 characters is too little.) > > I concur with the need to install an advanced version along special paths for > testing. Ideally that should be as simple as changing one HLQ. > > -- gil > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, send email to > [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
