Re: EXTPROC(*CL) and V6R1

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

Okay, I guess I see where you're going with this.  I'll try to word it 
more precisely for you:

a) If you call an RPG procedure that returns a 1-byte return value, and 
that procedure is called from C or CL, then you need to code 
extproc(*CL) (or extproc(*CWIDEN) or extproc(*CNOWIDEN)) on the 
prototype, and you need to recompile the module containing the procedure.

b) Any callers that will call that *copy* of that procedure, and are 
written in RPG, will need to use the new prototype, and therefore will 
need to be recompiled as well.

c) If you have programs/srvpgms that have the module bound by copy, and 
don't call it from C or CL procedures, you could potentially skip 
recompiling/rebinding those.  However, the NEXT time you want to change 
either the procedure with the issue, or the caller of that procedure, 
make sure you recompile/rebind both, as the source for the prototype 
will have changing (owing to point b, above) and therefore both the 
caller and the callee will need to be recompiled at the same time.


It gives me a headache just thinking about how you'd determine which 
programs need to be recompiled and which ones don't.  I'd just recompile 
the whole lot and be done with it.


David FOXWELL wrote:
> Although I still don't understand why the programs that are bound by
> copy would have to be recompiled. Wouldn't they continue to use the
> older version of the procedure? Then again, I don't know why the
> manager would not want to recompile all the programs using that
> procedure.