Re: Service Program Argument Optimization

[Steve Richter <stephenrichter@xxxxxxxxx>]

On Thu, 13 Jan 2005 12:32:46 -0500, Jon Paris <Jon.Paris@xxxxxxxxxxxxxx> wrote:
> >> It is interesting to me however because it invalidates my assumption of
> how the compiler passes parameters.
> 
> Chances are (based on my knowledge of the OPM compilers) that any variables
> used as parms have a pre-initialized pointer variable associated with them.
> Hence no need to evaluate and assign the address - it is already known.  The
> compiler simply passes the associated pointer.

But every variable is based on a pointer.  A static variable is based
on the pointer to static storage and a variable in a procedure is
based on the pointer to the call stack entry of that procedure.

> 
> Remember - we're not dealing with a C type scenario where few addresses are
> known at compile time.  Unless the variable is based or received as a
> parameter, the RPG compiler knows exactly where it will be in memory.

As much as I would like to argue with you Jon, I cant because I dont
know the facts. :)

One other factoid on this subject ....  setting a native pointer on
the as400, that is the 16 byte, full of capabilities pointer, is
actually a relatively slow performer compared to pointer operations on
other systems.  Setting a PASE pointer for instance.  So I was sure
passing by reference would be slower than passing by value.  But I was
wrong. Go figure.

-Steve