On Wed, 25 Mar 2020 13:28:12 +0000, Seymour J Metz wrote:
>"The MVSVAR('SYMDEF',string) function goes through REXX substitution for
>string first, the result of which must be a 1-8 character symbolic-name
>specifying the symbol that has been defined in the SYMDEF statement. Any other
>values including REXX delimiters might cause unpredictable results."
>
I see. But why!? is this an ill-conceived attempt to appease CLIST
programmers by supporting synthesized variable names?
is the "REXX substitution" identical to that which would be performed
by VALUE( string ); compound symbols resolved according to the
Symbolic rather than Direct convention, etc.?
o If not, the exact behavior should be specified.
o Even if so, a statement to that effect should appear.
(I suspect, hope actually, that this uses IRXEXCOM with the SHVSYFET option.)
>________________________________________
>From: Bruce Hewson
>Sent: Wednesday, March 25, 2020 5:43 AM
>
>In a REXX exec I was building I stumbled onto:-
>
>Say 'MVSVAR'("SYMDEF",'testing &sysname in &sysplex')
>
>which provides an unexpected result
>
>&'TESTING SYSA IN PLEX01'.
>
>The symbols &SYSNAME and &SYSPLEX were replaced.
>And, sadly, the whole lot was uppercased.
>
>Couldn't find this behaviour documented.
>
Was there no error indication? I'd expect one because of "must".
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
