ross@ghs.com said:
> > This was an n_long, not u_long, so u_int32_t is correct.
> 
> The reason for making this `unsigned long' is for compatibility with
> the historical arpa/inet.h, which will be included if
> !defined(_KERNEL) and !define(_STANDALONE). 

Sorry - this appeared to me like simply overlooked.
The n_long - ie 32-bit - has been there all the time and there
was no indication that it is necessary to change the interface.

btw, on my alpha/OSF1, inet_addr() gives an "in_addr_t" type which is
unsiged int, ie 32-bit, and they claim that it conforms to XPG4
in their manpage.
Comments from the standards experts?

best regards
Matthias