Re: setsockopt() compat issue

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Sun, Oct 19, 2008 at 12:57:59PM +0200, Johnny Billquist wrote:
> >The only thing that having int 64 bits solves is programs that assume
> >sizeof(int) == sizeof(long) keep working, while making life more difficult
> >in general because you are not going to have a primitive type for 16 or
> >32 bit numbers.
> 
> Huh? What happened to short now?

you have short be either 16 or 32 bits, and you miss the primitive type
for the other one.

> 
> Also, don't C99 actually mandate that [u]int{8|16|32|64}_t should be 
> defined.

it wasn't 10 years ago. So you had to introduce a non-standard primitive
type for 32bits values in a IPL64 world.

> 
> Programmers who actually rely on a int being 32 bits are idiots (sorry 
> if I offend someone, but it's a fact) and should be shot.

They don't rely on it being 32bits, they (did) rely on it being the same
lenght when the system runs in 32bits and 64bits mode.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--