On Tue, 22 Sep 2009 08:15:59 +0300, Binyamin Dissen <[email protected]> wrote:
>... >:>>I'd be quite satisfied with this _provided_ it was compatible >:>>with the existing CALL, LINK, ATTACH, etc. argument lists. >:>>I.e. on program entry R1 points to a fullword which points >:>>to a halfword containing the length of the PARMX followed >:>>by the PARMX string. >... >Actually the current interface does set the high order bit of the word >pointing to the parm to x'80'. All (almost all?) IBM utilities support >a three word plist, where the second word points to a list of >overriding DDNAMEs and >the third word points to a starting page number. I don't think any new arguments, counter-arguments, or proposed solutions are being raised here (and I'm certainly not about to), but the above comment adds a new twist. It confirms that the end-of-parms flag must be set for parms provided by JCL but also confirms that having both PARM and PARMX would be misconstrued by those programs expecting an optional list of overriding DDnames. This almost supports those saying that only one parm - PARM or PARMX - must be allowed. Almost. A program (foolishly) written with the assumption that it will always be invoked by via JCL with a 100-byte limit on parm length might (foolishly) keep the length in a single byte. It *might* get by with a max 255-byte parm but not a longer one. And when it abends the owner (possible a vendor), not the JCL "coder", is probably going to be called. Whatever technique used to pass a long parm is not going to be safely handled byt the old, standard parm passing scheme because a long parm passed that way will break old, standard programs. Some new technique that can be verified by new programs is needed or a lot of stable production jobs are going to die. I know some people see that as hobbling z/OS with severe limitations of the S/360 past, but I see maintaining compatability as a source of the great stability of z/OS and its ancestors. Thee is an obvious need for lifting the 100-byte limit on parms, but it needs to be done in a way that allows old programs to continue to work. Pat O'Keefe ---------------------------------------------------------------------- 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
