On Fri, 4 Aug 2006 18:09:35 -0700, Edward Jaffe <[EMAIL PROTECTED]> wrote: > > Ted MacNEIL wrote: > >> If you want to enter only uppercase data while editing JCL, simply press > >> your "Caps Lock" key > . In a modern 3270 emulator, it works a whole lot better than it did on the > old 3270 devices (on > ly alphabetics are capitalized; numerics remain numeric). It's no big deal. > >> > >> You're wrong! > >> It is a big deal! > >> ISPF will run CAPS off if there is one character in lower case in the > >> member. > >> > >> As a vendor, it may seem fine to you. > >> But, you should aim for the lowest common denominator. > >> > >> All JCL members should be in upper. > >> You, as a vendor, don't get a vote. > >> I've lost track of the attributions to Shane, TeD, (et. al.) Refer to the archives.
> Every vendor in every business is someone else's customer. And mainframe > ISVs are mainframe customers just like our customers are. We > install/maintain systems and software from ourselves, IBM, other ISVs, > etc. And, we definitely install our own software _way_ more often than > any customer will. > This cuts both ways. Familiarity breeds tunnel vision. I've been testing installation of a new product intensely using our installation JCL samples to the point where the operation has become almost a reflex. Accordingly, I can run through an installation without consulting the instructions; as a result the instructions are not validated until the reach (fortunately) a downstream tester. Worse yet, I've automated the process by constructing a script, which is _not_ suitable for dissemination to customers, to perform the "CHANGE ALL"s. (I tried to make it customer-worthy and failed miserably; my colleagues observed, correctly, it would be a net increase in complexity for the customers. I relented and abandoned it.) > process. Our goal is make it as easy as possible for us to install our > own software, and therefore easy for our customers to do the same. > As I note, these goals may not be parallel. And the remark (Shane's?) that if I need mixed case in, e.g. a HFS path name, I should INCLUDE similarly increases complexity. Then I'd need to instruct the customer to install the JCL in a particular library, and customize a JCLLIB statement in each member. I'm interested in the comment that mixed case should be rigorously avoided in JCL samples. My colleagues before me (Im following the convention, so far) quite deliberately used lower case in the "must change" strings. I assumed this would provide the clear benefit that if a customer inadvertently neglected a change a JCL error would result, preferable to a job's running amok. I might be persuaded to change this, but not in time for the release we're working on. I'll listen to suggestions. I wish I could automate the process more, but I fear that wrappers, scripts, and utilities to improve on an SMP/E installation are likely only to be an Aggravator. -- gil -- StorageTek INFORMATION made POWERFUL ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
