Hi,

I have an array of Byte which I have loaded from a file.

I want to process a 100 byte section of the array which starts at offset 32 in my array.

My function (or procedure) is as follows:

procedure decompress_r(cbuf: array of byte);
var
  len,i: integer;
begin
  i:=1;
  if (cbuf[0]and $F0)=$20 then
  begin
    Inc(i,2);
    len:=cbuf[i]and $7;
  end;
  ...
end;

I call my procedure by

decompress_r(&buf[32]);

It seems to work. the first element in the array is 0x20, but the rest of the array is garbage.

The array sent is:
buf[32]=32
buf[33]=0
buf[34]=0
buf[35]=7
buf[36]=0

The array received is
cbuf[0]=32
cbuf[1]=27
cbuf[2]=255
cbuf[3]=247
cbuf[4]=255

Can anyone tell me the correct way to pass the portion of the array to the function?

Many thanks.

Dave

Did you really write "&"? The address operator is "@", but this won't make your code work anyway. If you did indeed write "&" (which is for escaping keywords) you do indeed only pass the 32nd element to your function.

Dynamic arrays are internally pointers to an array (more or less). So if you would use "@buf[32]" you would be passing the address of one element to a function that takes an array. If you know the amount of bytes you want to copy then you should use "Copy(buf, 32, count)" as argument for decompress_r or if you don't know the amount you should change your function from "array of Byte" to "PByte" and pass the address of the 32nd byte using "@buf[32]". The remaining part of the function can stay the same.

Regards,
Sven

There is no looping construct (while or repeat) in your decompress_r procedure.

Howard

If you know the amount of bytes you

want to copy then you should use "Copy(buf, 32, count)" as argument for decompress_r or if you don't know the amount you should change your function from "array of Byte" to "PByte" and pass the address of the 32nd byte using "@buf[32]"

Or use the array slicing construct:

func(buf[32 .. <Last index of the array>]);
reference: http://www.freepascal.org/docs-html/ref/refsu59.html

And it's still true: even I as a compiler developer can learn something new as I didn't know yet that this was possible.

Regards,
Sven

Sven Barth wrote

Did you really write "&"? The address operator is "@", but this won't make your code work anyway. If you did indeed write "&" (which is for escaping keywords) you do indeed only pass the 32nd element to your function.

:blush:

I did indeed. That would explain it!

Thank you, Sven, that will help a lot.

Many thanks,

Dave

:-) Wasn't there something in PL/1 that allowed a skewed slice of a 2-D array to be passed as a 1-D parameter?

The delphi compatible way is slice, but ti was only documented after the last release:

http://bugs.freepascal.org/view.php?id=22681

I did know about Slice, but not about the possiblity to use ranges directly (and now I know what the corresponding code in the parser is for :P )

Regards,
Sven

Afaik it was one of those "FPC must find its own way, and not slavishly follow Delphi" features. Usually orphans at birth.