Re: Wrap an RPG Procedure with an SQL External Procedure

[Buck <kc2hiz@xxxxxxxxx>]

Joe Pluta wrote:
> We'd have to ask someone more knowledgeable about the compiler to see if the
> following idea is safe, but you MIGHT be able to take the address of the
> return value and compare it to NULL.  If it's null, don't set the return
> value.
>
> It's not perfect, but if it's safe, it beats the double maintenance of a
> wrapper stub.

I have been able to corrupt storage I don't own by doing something 
similar.

CALL SOMEPGM /* no parms on call by accident */

SOMEPGM:
*entry plist
       parm  returnCode
       eval  returnCode = *blanks

This code has run without getting a MCH3601.  Sometimes there's a 
pointer in just the right spot in memory and BLAM!

Never done that with a function/subprocedure call because RPG lets me 
discard the return value (callp someProc vs eval return = someProc). 
Apparently the database manager isn't so forgiving.  It would be an 
interesting test to see if one *always* got a NULL from the database 
manager.
  --buck