Just a little idea about more portable code with less conditional compilation...

In >90% cases little-endian / big-endian conversions are used to access external data with predefined format.
In such cases it's more clear and portable to specify byte order explicitly as a type attribute.
By default everything has byte order native for the target platform.
This attribute can be changed locally if necessary:

// something like this:

big_endian  uint32  *p; // pointer to big-endian data

little_endian
struct  SomeData
{
    int32  a_little-endian;
    big_endian  int16  b_big-endian;
    uint16  c_little-endian;
...
}

Serge K wrote:

> In >90% cases little-endian / big-endian conversions are used to access external data with predefined format.
> In such cases it's more clear and portable to specify byte order explicitly as a type attribute.
> By default everything has byte order native for the target platform.
> This attribute can be changed locally if necessary:
> 
> // something like this:
> 
> big_endian  uint32  *p; // pointer to big-endian data
> 
> little_endian
> struct  SomeData
> {
>     int32  a_little-endian;
>     big_endian  int16  b_big-endian;
>     uint16  c_little-endian;
> ...
> }

The problem I see with this is that if you do any kind of manipulation of b_big-endian on a little-endian machine you'll end up doing a lot of endian conversion unnecessarily.

By specifying it explicitly you can leave all the endian conversion until it really has to be done.



-- 
Chris Friesen                    | MailStop: 043/33/F10
Nortel Networks                  | work: (613) 765-0557
3500 Carling Avenue              | fax:  (613) 765-2986
Nepean, ON K2H 8E9 Canada        | email: cfriesen@nortelnetworks.com