[...]
> Everyone who uses <linux/foo.h> is going to be bitten.
Yes, and (as you claim) userland programs work fine because the userlevel
interfaces haven't changed. And the kernel internals are nobody's business
(except Linus and his gang, that is :)
BUT: There is userland, kernel sea, and systems shore. By the last I mean
programs like mount(8), fsck(8), ifconfig(8), hwclock(8), hdparm(8),
modprobe(8), ipfwadm(8)/ipchains(8). Sure, they should work with 2.0.xx
and 2.1.xxx if at all possible (sometimes it isn't, see last example). It
is not totally unreasonable for shore programs to #include <linux/...> or
even #include <asm/...>. Yes, their authors will get bitten by kernel
changes. Just keep a snakebite kit handy ;-)
For the interface between the kernel and the outside world (i.e., mostly
libc) it is critcal having a clean access to the internals, with truly
internal cruft hidden by __KERNEL__ or whatever.
Pure user programs I agree totally with you.
-- Horst von Brand vonbrand@sleipnir.valparaiso.cl Casilla 9G, Viņa del Mar, Chile +56 32 672616- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/