Re: Prototyped Procedures

[Bruce <bruce@xxxxxxxxxxx>]

This isn't a response to anybody in general, just my thoughts. The 
thread has gotten a little crazy to follow.

We use both methods where I work, and I've use a variation of the 
/define version in a few instances too.

Almost all of our prototypes are in a single source member.  Prototypes 
are preceded with a p then the first nine characters of the function 
name.  We normally don't define work variables or ds in the /copy member 
unless for when we're prototyping external programs that needs i/o 
fields.  Then we prefix with something then an underscore (addr_Zip).

There's one reason I like the /define version in that it allows for the 
full function name to be used.  But, I think it is horrible to force the 
app developer to have three lines to do one thing.  The black box 
programmer needs to do three line for one thing.

A variation that I use is to auto define depending on what has already 
been /included. I use this method, probably the wrong way, instead of 
multiple module binding.  I pull in the source the old fashion way to 
each prog that uses it.  I present it here as a variation.
For example:

/if NOT defined(MyFunction_Prototype)
/define MyFunction_Prototype
 MyFunction   PR
 ...
/eof
/else
/undefine MyFunction_Prototype
 MyFunction   PI
 ...
/endif

Then in the applications I have a /copy in the D specs and one in the P 
spec area without having any /define.

My preference, if anybody really cares at this point in the thread, is 
individual /copy members due to an attempt at self documentation.
My wish would be to have 32 char member names so we could use a member 
name that more matches the function name.

Take it or leave it...
Bruce in AR