Re: Why is there a "name" on the procedure interface?

On 2019-04-26 6:51 PM, Stuart Rowe wrote:

I always wondered why there is a "name" on a procedure interface. ...
Can anyone show an example of where a non-*N name is absolutely necessary
(and meaningful)?

In most cases, the name of the PI is the same as the name of the procedure, so *N is fine for the PI.

The only time you need a name is for the cycle-main procedure where you have a prototype in a /COPY file. The compiler needs the name on the PI to know which prototype to use.

In the following example, the "incorrect program"has *N on the DCL-PI, so the compiler doesn't know that it had to match the MYPGM prototype that was copied in, and it wouldn't detect that the parameter for the PI doesn't match the parameter for the PR.

The "correct program" has a name for the PI, so the compiler can match the PI with the PR and detect the error.

/COPY file MYPGM_PR:
dcl-pi mypgm;
p1 char(10) const;
end-pi;

Incorrect cycle program with *N on the PI:
/copy MYPGM_PR
dcl-pi *n;
p1 char(20) const; // compiler doesn't flag parm mismatch
end-pi;

Correct cycle program with a name on the PI:
/copy MYPGM_PR
dcl-pi mypgm;
p1 char(20) const; // compiler flags parm mismatch
end-pi;