BS> Could someone please give me a advice of what I should keep in
 BS> mind, then I won't to write Endian-less software?

 JB> in the past

 JB> The most straight-forward way was when reading and 
 JB> writiing integers, floats
 JB> etc process them a character at a time.

 JB> another way is to make a function that reverses the bytes in the number
 JB> and use that before writing and after reading...

 JB> Then you arrange things so that when the edianness of the data file
 JB> matches that of the CPU that then the byte-shuffle routine doesn't
 JB> run.

Yes,

 JB> there's two ways to do that too.

 JB>    One is to always use the same endian-ness in the data file and if
 JB>    compiling on correctly endian hardware #define an empty macro to
 JB>    leave the bytes alone

 JB>    some environments (eg POSIX) provide htonl() etc which does this for
 JB>    you.

Yes I did see, and now I understand what host byte order and network byte order
is :)

 JB>    The other is to have the software detect the endianness of the data
 JB>    file and set a flag which youv'e implemented to switch the
 JB>    byte-swapping on or off at runtime.

Like I showed from the husky project?

Bo


--- Maximus/UNIX 3.03b