Not really. Well, at least the only difference I've really noticed
is that I can do "make modules" in my kernel source tree, install them
and have it whinge and complain about some symbol mismatch or whatever
and refuse to load them. I'm sure there is a good reason for it but I
don't think it is being done the right way if I can build modules from
the same source tree as the kernel and not have them work at all. Maybe
I'm doing something wrong which is very simple, who knows ? (Hence the
kernel daemon which loads them has questionable value to me).
Maybe the *BSD's don't have automatic loading of LKM's, but at least if
I build one from the same tree as the kernel it doesn't tell me I can't
load it because of a version label problem!
> Regarding the /usr/include/* differences -
> I hope you don't want to introduce mbufs in Linux ;) glibc goes further
> into the Single Unix compliance direction, but even libc5 isn't that bad.
> I usually find all the BSD extensions much more annyoing during ports.
Heck, I'd settle for a *sane* /usr/include/netinet and /usr/include/net
on Linux without the mbuf's. It is possible but it'd mean using BSD
include files/header styles (which is too much to expect of a lot of
Linux die-hards here). Everything is different and different to every
other Unix (who all inherited these files at least, from BSD). How
completely fucked is that ?