Re: Using printf() function with RPG...Where did the results go?

[Barbara Morris <bmorris@xxxxxxxxxx>]

"mlazarus@xxxxxxxx" wrote:
> 
> >By the way, make sure you have at least one OPTIONS(*NOPASS) parameter
> in your prototype even if you always want to pass all the parameters.
> Otherwise some of your parameters won't be passed correctly.
> 
>  Can you explain why this would be?  I sems to be an odd requirement if all
> parms are being passed.

If a C prototype has ... (ellipsis), the RPG prototype has to have at
least one OPTIONS(*NOPASS) parameter - actually it probably should have
OPTIONS(*NOPASS) on the first RPG parameter that isn't listed in the C
prototype.  A ... function is a "var-args" function, which means the
parameters following the last declared parameter must be addressable
contiguously.

I believe that this rule about the parameters being addressable in
storage applies only to "primitive" parameter types (the numeric types
including single-byte char, and pointer).  I don't know what the system
does with other types passed by value like RPG's long character types.

If OPTIONS(*NOPASS) isn't coded on any parameters, the system is free to
pass some parameters in registers, but printf would not be able to see
those register parameters.

The most recent experience I've had with this was for the open()
function - the fourth parameter was not being read correctly.

If this was a surprise to you, there might be more C/RPG things that are
useful to know.  This site http://www.opensource400.org/callc.html has
most of what I know about the various gotchas.