From: x@x.com (Alex Grigny de Castro)
Subject: Re: Mac numeric float format?
Date: 25 Sep 1999 00:00:00 GMT
Message-ID: <37ece1aa.619650269@news.A2000.nl>
Content-Transfer-Encoding: 7bit
References: <37ec26f4.571852019@news.A2000.nl> <7shlko$b7$2@bgtnsc03.worldnet.att.net>
Content-Type: text/plain; charset=us-ascii
X-Complaints-To: abuse@a2000.nl
X-Trace: weber.a2000.nl 938271717 23931 24.132.1.94 (25 Sep 1999 15:01:57 GMT)
Organization: A2000 Kabeltelevisie en Telecommunicatie
Mime-Version: 1.0
NNTP-Posting-Date: 25 Sep 1999 15:01:57 GMT
Newsgroups: comp.lang.pascal.delphi.misc

Thanks a lot, Earl, it seems to work perfectly !  The code I used was:

In the stream reading proc:
var sg: Single;
//...
// Joint.pw is a Single

      BytesRead := Stream1.Read(sg, SizeOf(sg));
      Joint.pw := zReverse(sg);
//...

The zReverse overload functions:

function zReverse(a: Single): Single;
type pint = ^Integer;
type psng = ^Single;
var p: Pointer;
    pig: pint;
    psg: psng;
    ig : Integer;
begin
   p := @a;
   pig := pint(p);
   ig := pig^;
   ig := zReverse(ig);
   p := @ig;
   psg := psng(p);
   Result := psg^;
end;

function ZReverse(a: Integer): Integer;
begin
   Result := ($ff000000 and (a shl 24))
          or ($00ff0000 and (a shl 8))
          or ($0000ff00 and (a shr 8))
          or ($000000ff and (a shr 24));
end;

You made my day :)

Alex Grigny.


On Sat, 25 Sep 1999 00:20:08 -0500, "Earl F. Glynn"
<EarlGlynn> wrote:

>Alex Grigny de Castro <x@x.com> wrote in message
>news:37ec26f4.571852019@news.A2000.nl...
>> I'm trying to read binary files produces by Macintosh software, which
>> contain floating point and integers in their format. I've sorted out
>> the integer bit - just reverse the order of the bytes - but am
>> puzzeled by the floating point.  They are blocks of 16 bytes,
>> presumably containing 4 numbers (quadrions for a 3d transformation).
>> Does anyone know how to do this, or know the Mac specs for this?
>
>My guess is that Mac floats are "big endian" IEEE values.  So
>the values can be 4 bytes (single) or 8 bytes (double) long.
>You need to reverse the floats, just like the integers, to convert from
>the Mac's big endian to the PC's little endian format.  Here's
>some info about how to do this in Delphi:
>http://www.efg2.com/Lab/OtherProjects/Endian.htm
>
>For more info about IEEE floating point values, see Section B of
>http://www.efg2.com/Lab/Library/Delphi/MathInfo
>___
>efg
>
>Earl F. Glynn
>Overland Park, KS  USA
>
>efg's Computer Lab:  http://www.efg2.com/Lab