On Tue, 30 Aug 2016 19:07:09 -0500, Tom Marchant wrote: > >Perhaps even better, and to satisfy Gil's requirement, define a system symbol, >perhaps &NULL, with a substitution value that is null, and use that with >SYMBOLICRELATE to SYS1.FOOBAR&NULL > Does that restrict the length of related DSNs to 39 characters rather than 44 because a 5-character name is replaced by a 0-character value? Of course a thoughtful design would allow the SYMBOLICRELATEd DSN to be comfortably longer then 44 characters because symbol resolution may reduce that length. Somehow I don't trust IBM designers to have been so thoughtful. For example, ISPF DSLIST restricts the name entered to be at most 44 characters regardless. I'd be delighted to be surprised.
>or some such. I suspect that you don't even have to define the system symbol, >that if &NULL is not defined, you will get the same result, but I can't test >until >tomorrow. > It depends on whether catalog is like many UNIX utilities that take an undefined symbol to be null, or like JCL, Rexx, ... that take the value of an undefined symbol to be its name. I think a more cautious design of JCL would have reported an error on use of an undefined symbol (like SIGNAL ON NOVALUE). But I picked up many bad habits decades ago from a wizard/mentor who delighted in exploiting unintuitive idiosyncrasies of JCL. Much of this this can be tested without altering the master cat (that was only the OP's example), but any two user catalogs. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
