On Wed, Dec 25, 2013 at 09:44:39AM +1100, Jason White wrote:
Matthew Cengia <mattcen(a)gmail.com> wrote:
I loathe the idea of systemd as much as the next
guy, and basically
agree with everything you said in your previous email,
I don't loathe the idea of systemd; in fact I think it's an
interesting and useful piece of work, which is why it's seeing
adoption from multiple distributions (Arch, Fedora, OpenSUSE, possibly
There have always been different init systems (System V, BSD, etc.) in
use, and there's nothing to suggest the situation is about to change,
hence I don't agree with the "monoculture" claims.
if it was *just* an init system, then it wouldn't be a monolithic
monstrosity resulting in a mono-culture. but it's also absorbing
syslogging, cron, udev, console and login handling, automount, and
everything else in sight.
an init system doesn't need to do all those things. it just needs to
handle starting up daemons/services in the right order at boot time, and
killing them later on shutdown or reboot.
at the moment, each one of those system functions is provided by a
separate, independent program - and each one can be replaced with an
alternative or competing implementation *WITHOUT* fucking up the rest of
the system. and we have *all* benefitted from that modularity over the
years. I want to keep benefitting from that in the future.
with systemd, that will no longer be possible. it's all or nothing.
systemd's monolithic nature is antithetical to the unix "small tools"
philosophy of "do one thing only but do it extremely well" that has
served us all so well for decades. systemd absorbs everything and does a
just barely good enough job for each of them before moving on to absorb
something else (i.e. jack-of-all-trades, master-of-none) and the end
result isn't anything like unix. it's not unix and it isn't linux any
more, which is why i've taken to calling it lennartix.
It's true that the developers of systemd are keen
to promote their
work, but that's nothing new in software generally or free/open-source
promotion is one thing. absorbing every low-level system service whether
it's related to init or not is quite another. building unneccessary
interdepencies that require end-users to switch to systemd is not mere
"promotion", it's RH's version of the microsoft embrace and extend
tactic, but in this case it's absorb and monopolise.
having multiple corporations contribute to and co-operate and compete
in linux development is a good thing. having one corporation end up
controlling the low-level userland AND the desktop environments of all,
or nearly-all, distros would be an absolute, unmitigated disaster.
Furthermore, I'm not suggesting to anyone that
they should (or shouldn't) use systemd; that's an issue to consider,
among others, in choosing a distribution.
it's more than just in choosing a distro - there are growing
interdependencies between systemd and gnome (and, unfortunately, other
desktop/window managers - even xfce). it's not quite there yet, but the
future is clear: if you want to run gnome or most other desktops, you
will have no choice but to run systemd (or, if you're lucky, put up
with running your desktop environment in a second-rate "unsupported"
systemd-less "degraded mode")
when you *can't* choose to run something else - whether it's a syslogd
or a crond or your desktop - because of these interdependencies, *THAT*
is when you get a complete mono-culture.
craig sanders <cas(a)taz.net.au>