FreeTDS Development Group

["James K. Lowden" <jklowden AT schemamania.org>]

On Wed, 7 Aug 2002 23:13:11 +0100, "Thompson, Bill D (London)"
<ThompBil AT exchange.uk.ml.com> wrote:

> On a big endian machine the two 4 byte integers, strung together , can
> be treated as a single entity.
> On little endian of course they cannot. 

Of course?  I guess I don't understand what "strung together" means.  :) 
We can do 64-bit integer math.  Does that count?  

> Assuming we dont change tds_process_row, 

But we probably should ...

> and the money value comes down
> unchanged the bytes would have to be moved as follows ( I reckon, and it
> IS late here )
> 
> Wire: 
> 
> 1 2 3 4   5 6 7 8
> 
> To:
> 
> 5 6 7 8   1 2 3 4 
> 
> does that make sense ?

Thanks, Bill.  It makes sense, except that I was expecting:

8 7 6 5 4 3 2 1

(assuming each digit represents one byte).  That's what you get with:

        char buf[] = "12345678";
        int i = tds_swap_bytes( buf, 8 );
        printf( "%d bytes swapped: \"%s\"\n", i, buf );

I also experimented with ntohl and htonl, program below.  The output:

$ gcc -g  -o swaptest swaptest.c 
$ ./swaptest 
8 bytes input:    "12345678"
8 bytes swapped:  "87654321"
8 bytes reverted: "12345678"

ntohl,  input: "1234"
ntohl, output: "4321"
htonl, output: "1234"

I wonder what "htonll" would do?  

I'll mangle a few and see how it goes.