In article <1fma705.ovpquf1mdrc0eN%dempson@actrix.gen.nz>,
 dempson@actrix.gen.nz (David Empson) wrote:

> Yves McDonald <""yves.mcdonald\"@NO SPAM sympatico.ca> wrote:
> 
> > Ok, the situation so far is:
> > a) i/o address is located in first 64 kb memory segment in GS memory map
> 
> They are also in the $01Cxxx, $E0xxxx and $E1xxxx areas, but $00Cxxx is
> the easiest area to deal with it because the numbers will be the same as
> the (unsigned version of the) Applesoft BASIC PEEK and POKE addresses.
> 
> > b) a byte pointer could be used to access I/O address, thanks to ORCA
> > Pascal pointer arithmetics.
> 
> This assumes that accessing memory through a byte pointer will actually
> perform an 8-bit memory access.  ORCA/Pascal might be taking a shortcut
> and using 16-bit mode even for a nominal 8-bit read operation, and
> discarding the high order byte of the result (this avoids two CPU mode
> switches).  It must switch to 8-bit mode for writing via a byte pointer.
> 
> I haven't use ORCA/Pascal for this sort of low level operation, but
> probably tried it using ORCA/C, which has similar low-level code, but
> don't recall whether I ever did direct access to I/O locations.
> 
<snip>

Out of curiosity I tried this trivial program to see if I could click 
the speaker in Bernie to the Rescue:

 + - + - + Cut click.pas + - + - +

{$keep 'click'}
program click(output);
 
var
  s: byte;
  p: ^byte;
  i: integer;
 
begin
  p:= Pointer($0c030);
  for i:= 1 to 27000 do begin
     s:= p^;
  end;
end.

 + - + - + End click.pas + - + - +


It caused Bernie to click the speaker. Cool, I learned something else 
new today :o) It's been so long since I wrote a Pascal program I hope I 
didn't do something wrong ;o)