My point originally was the development teams do not sit down with the documentation and go through it from scratch. The testers have a test suite called "installation tests", they just press a button. You do not want the development teams to do this work because they know too much about the product. You want someone who has never done it before, but is skilled on the platform.
I expect a lot of testing is done with security disabled. The default should be all testing is done with tight security. The exception is no security. An example is setting up certificates and keyrings. Once you have done it once you just reuse it. I found the most recent (and strongest) cipher specs do not work with some products. On Mon, 29 Nov 2021 at 00:41, Charles Mills <[email protected]> wrote: > Not what I am suggesting. I am not suggesting that coders of a General > Ledger system also be accountants who use it. I am suggesting it at the > corporate level. Use in your enterprise what you sell to your customers. > > Not in every case of course. If IBM brought out a GL system targeted at > small businesses you would not expect IBM to use it internally. > > But you would certainly want developers to be using the same documentation > system that customers were expected to use. > > Charles > > > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[email protected]] On > Behalf Of Paul Gilmartin > Sent: Sunday, November 28, 2021 11:41 AM > To: [email protected] > Subject: Re: IBM RFE's Broken > > On Sun, 28 Nov 2021 11:08:11 -0800, Charles Mills wrote: > > >> Oh we never use the product > > > >It is a real bad sign when any software company does not eat its own dog > food. > > > You're requiring a person who develops, tests, uses, and documents > the product; an unlikely combination of talents. Particularly, most coders > are not good tech writers or are unable to view the product with an > unbiased eye. > > It's a boon when a compiler is implemented in its own source language. > > >-----Original Message----- > >From: IBM Mainframe Discussion List [mailto:[email protected]] On > Behalf Of Colin Paice > >Sent: Sunday, November 28, 2021 10:45 AM > > ... > >Many a time I've asked myself, "why is this solution so complex". It is a > >simple problem. Simple problems need simple solutions. I should not have > >to configure a 7,000 line rexx program! > > > (Cite? But I don't doubt it. And GNU Autoconf?) > > >I remember one tester saying "Oh we never use the product - we have > >automation which runs all of our tests for us. > > > But such a test suite should include extreme cases. > > I once reported an error in the SDSF API for SYSOUT,LRECL=32753. > Support (grudgingly?) agreed to fix it, but only for LRECL>=32753, lest > some customer, somewhere, had come to depend on the brokenness, > perhaps only by compensating for it. > > Compatibility can be a curse. > > -- 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 > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
