Steve Litt <slitt@???> writes:
> On Wed, 7 Oct 2015 11:58:25 -0300
> Emiliano Marini <emilianomarini82@???> wrote:
>
>> This is a great compilation about systemd from softpanorama.org
>>
>> http://www.softpanorama.org/Commercial_linuxes/Startup_and_shutdown/systemd.shtml
>
> I'd say this document provides a nice strawman for Red Hat to knock
> down. Like so many others, the author lets Redhat and Lennart bring the
> war to us:
>
> ==========================================================
> "I would not deny that improvements in init are possible."
> ==========================================================
>
> Not a mention of Epoch. Not a word of s6, runit, nosh, perp, busybox
> init, or even Suckless Init + daemontools-encore + LittKit. He fights
> the war where we're the weakest: the 30+ year old sysvinit.

How about not comparing apples to oranges? The sysvinit program does
contain some process management code but its virtually unused,
presumably, because it was supposed to be compatible to the actual
System V init. The code could be improved by removing that. Apart from
that, the only thing it really does is 'execute certain, configurable
commands in response to system state change requests'. That's general
enough to implement anything based on it, including systemd-style
process management (eg, SMF cooperates with init but doesn't replace
it. Neither does perp, at least according to the README).

The process management support available in 'common Linux distributions'
prior to systemd can surely be improved in many ways. But none of them
requires dbus to function or means the messieurs Poettering and Sievers
had to go onto a "rewrite whatever doesn't run away quickly enough"
spree the result of which everybody else is supposed to accept with
graceful humility no matter how it looks like and how useless/
contraproductive it happens to be for him.