On Mon, 22 Feb 2016 14:49:20 +0000
Simon Hobson <linux@???> wrote:

> In my experience, sometimes
> you just have to accept that some edge cases don't work very well -
> but it's better to have to fudge around those than to make the entire
> world suboptimal to cater for them.

Read the preceding sentence. Isn't that the philosophy we sans-systemd
people live by every day?

Apparently systemd works, from what I read. Apparently it works for
everybody, from the dumbest of the dumb to the smartest of the smart,
in all use reasonable use cases. But at what cost?

Every edge case has its own little connector within systemd. There is
layer upon layer upon layer of airbags and safety shutoffs and
auto-configurators within systemd. Instead of merely providing a way to
read the root partition in order to get to /etc, /bin and /sbin,
systemd provides a jungle-like initramfs and actually interacts with it
over and over again.

When we sans-systemd people find a use case that doesn't work out of
the box, the solution is usually a couple shellscripts away: Those
shellscripts being simple because our systems are simple. When the
systemd crowd finds something that doesn't work (admittedly, this is a
rarer occurrence), they must be doing something wrong.

I like automatic. I like "just works". But not when it buries the
mechanism in a Rube Goldberg machine of official workarounds. Or, in
the words of a very observant person:

=========================================================
sometimes you just have to accept that some edge cases don't work very
well - but it's better to have to fudge around those than to make the
entire world suboptimal to cater for them.
=========================================================

SteveT

Steve Litt
February 2016 featured book: The Key to Everyday Excellence
http://www.troubleshooters.com/key