From: "DeHackEd" <notgiven AT out DOT com>
Newsgroups: comp.os.msdos.djgpp
References: <36630cd2 DOT 17414449 AT news DOT extension DOT net>
Subject: Re: Strange behavior of fread
Lines: 72
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Message-ID: <ZYD82.2704$Z72.846219@tundra.ops.attcanada.net>
Date: Mon, 30 Nov 1998 21:18:49 GMT
NNTP-Posting-Host: 142.194.222.233
NNTP-Posting-Date: Mon, 30 Nov 1998 21:18:49 GMT
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Probably a processor processing the data differently. If you know this is
happening, try reading in 1 byte at a type and << the appropriate byte 8 places.
Or you could try something like this:

inline short byteflip(short value)
{
    return value >> 8 | ((value&0xff) << 8);
}

I doubt it would make a difference, but try some optimizations and -mpentium
--
"DeHackEd"

My Email address in the header is fake (spammers). Email me at this address:

http://www.geocities.com/cgi-bin/homestead/mail.pl?dehacked
dalitz wrote in message <36630cd2 DOT 17414449 AT news DOT extension DOT net>...
>Hello folks,
>I observed an unexpected behavior of fread.
>It seems to me that the order of two subsequent bytes
>is swapped in the buffer when the fread function reads items of
>two bytes (short int). I really don't understand what's going
>on.
>
>Let me give you an example:
>
>...
>long int offset;
>short int *value;
>...
>ifile = fopen("binary.dat","rb");
>ofile = fopen("text.dat","w");
>...
>offset = 8*320;
>for (i=0; i<=160; i++){
>    fseek (ifile, offset+2*i, SEEK_SET);
>    fread (value,  sizeof(short int), 1, ifile);
> fprintf(ofile,"Offset:%d\t%d\n",offset+2*i,*value);}
>...
>
>The hex dump of binary.dat looks like that:
>
>Offset   Data
>...           ...
>000a90  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00
>000aa0  00 02 00 04 00 03 04 88  35 63 00 02 00 00 00 00
>000ab0  00 01 00 C8 00 8C 00 01  00 13 00 04 00 0A 00 1C
>...           ...
>
>This gives the following output in text.dat:
>
>...
>Offset:2720 512         this is 02 00 instead of 00 02
>Offset:2722 1024       this is 04 00 instead of 00 04
>Offset:2724 768         this is 03 00 instead of 00 03
>Offset:2726 -30716    this is not 04 88 but may be the result of
>88 04
>Offset:2728 25397     this is 63 35 instead of 35 36
>...
>
>I tried that with VC++ and djgpp on both Win NT and Win 95. The
>results were all the same.
>binary.dat was downloaded from an HP-UX 9.0x server an was created
>by some Fortran Code.
>
>Does anybody know what's going on here?
>I would appreciate any comment or idea. Thank you in advance.
>
>Dalitz.
>