Re: *ENTRY PLIST to free-form procedure interface

On 2017-10-10 5:26 PM, Dan wrote:

Thanks Joe,

Your assumption is what I presumed, but wasn't sure if there were other
instances where a procedure could have more than one procedure interface.

I'm trying to determine a good standard for our "*entry plist" procedure
interfaces, since we're fairly new at using them here. Personally, I think
I'd like to name these procedure interfaces Entry_Plist. For example:

dcl-pi Entry_Plist;
Parm1 char( 1);
Parm2 char( 1);
Parm3 char( 1);
end-pi;

If you are (now or ever) going to use prototyped calls to call your programs, then you'll need a different name for each prototype. And the procedure interface has to have the same name as the prototype to allow the compiler to check the parameters for consistency. Adding "*ENTRY PLIST" in a comment is a good way of letting people find the PI. Probably better than calling it Entry_Plist, since people would probably be searching for '*ENTRY', not just 'ENTRY'.

I recommend getting in the habit of putting the prototype in a /COPY file, and copying it into the program itself and any module that calls the program.

The only time I would have a program with an unnamed PI for a cycle-main program would be for something like a command-processing program that isn't ever intended to be called from another program.

For procedures that are exported from the module, the same thing applies: put the prototype in a copy file, and copy it into the module with the procedure and all calling modules. But you don't need to name the procedure interface in that case; you can just use *N because the name of the PI gets picked up from the name of the procedure.

For procedures that aren't exported from the module, a prototype isn't needed at all.