Re: Checking for missing parameters in CL

David,

I read this thread, but wasn't sure that anyone explained it completely, so...

If we're talking about *OMIT, what you're doing should work. Meaning: The caller actually passes *OMIT or passes a parameter with a null address.

In addition to using pointers, you can also call the Test of Omitted Argument (CEETSTA) API, which basically does the same thing, but you do an API call vs. coding a pointer. This worked before we had pointers in CL.

However, you should not rely on either of these if it's possible that the parameter won't be passed at all. I've seen this mistake too many times! It "seems" to work, because as long as what happens to be in the memory used for the parameter isn't a valid pointer, it'll register as "null", so the parameter checking seems to work this way, but as soon as whatever happens to be in that memory is a valid pointer, it will not. For example, if you call it twice, once with the extra parameter, and immediately afterwards without the extra parameter, the parameter count (RPG %PARMS) will show that the parameter is missing, but the extra parameter won't be null because the memory is still set to the pointer from the previous call, so it'll look like the extra parameters are passed.

I've even seen that mistake in commercial software, sadly.

Also, if it is a procedure call (vs a *PGM call) you can't even assume that %PARMS will be set, frustratingly, because %PARMS is part of the operational descriptor. But, since CL doesn't yet have a %PARMS, this is probably a discussion for another time. In a *PGM call, however, %PARMS will always be set properly.

-SK


On 1/31/2018 10:38 AM, David Gibbs wrote:

Folks:

I just discovered this and thought I would share ... not sure if it's already been discussed or if it's common knowledge. :)

I have a CL Bound program that I wanted to have an optional parameter.

I know I could create a command for the CL and the default for a unspecified parameter would, but this wasn't warranted for the particular change I was making.

I know, in RPG, you can specify *OMIT on parameters that have the OPTION(*OMIT) modifier ... and you check to see if the address of the parameter is null.

So I tried something similar in CL...

pgm (&parm1)

dcl &parm1 *char (10)

dcl &p1addr *ptr

chgvar &p1addr (%address(&parm1))

if (&p1addr = *null) then(do)
   /* do something if parameter wasn't specified */
enddo
else do
   /* do something else if the parameter was specified */
endpgm

So far it seems to work as I expect.

david

---
This email has been checked for viruses by AVG.
http://www.avg.com