Re: Installing an Alternative Init?
On Tue 11 Nov 2014 at 11:38:19 -0500, The Wanderer wrote:
> On 11/11/2014 at 10:54 AM, Brian wrote:
>
> > systemd is the default init system. That means everyone gets it.
>
> No - that only means that everyone gets it by default, not necessarily
> that everyone gets it.
Everyone gets it. Not everyone boots with it. Not everyone who boots
first time with it gets to use it on subsequent boots.
> Unfortunately, no one seems interested in recognizing that people *DO
> NOT AGREE* about what the word "default" means (or should mean) in the
> context of "the default init system", or in having a discussion about
> what it should mean - or even in figuring out what each other do mean by
> that term, and possibly finding other ways to describe those meanings so
> that the ambiguity goes away.
>
> > You can only have one init system as PID 1, so that means changing to
> > an alternative involves removing systemd first.
>
> Only if systemd is already installed as PID 1, which is precisely what
> the disagreement is about.
>
> You subscribe to a meaning of "default" which assumes that systemd must
> necessarily get installed as PID 1 before anything else happens. That's
> also what the current state of what actually happens is.
>
> Other people subscribe to a meaning of "default" which, e.g., assumes
> only that systemd will get installed as PID 1 unless some action is
> taken to prevent it from getting so installed. That seems like an
> entirely reasonable interpretation, at least to me.
>
> It looks to me like you're assuming the consequent - building your
> argument on the assumption that what your opponent is arguing against is
> the truth. That's not really a good way to make progress in any discussion.
Isn't the installation of systemd as PID 1 from, for example, a netinst
image a fact? In all this discussion no-one has disputed this.
> > "Clean" install is a bogus target. There is not a single technical
> > advantage in pursuing it as a feature to add to d-i. Changing the
> > init system within the package management framework works and has no
> > disadvantages.
>
> At the very least, it has the minor disadvantage of wasting resources
> (time, CPU power, write cycles, et cetera) on installing the non-desired
> package to begin with.
>
> Other disadvantages may be more a matter of opinion, but that one at
> least does exist, however negligible it may arguably be.
I'd like to say it is negligible and merely a consequence of using the
package management system as intended.
> (Hmm. There may be a parallel here; many of those objecting to systemd
> do so on the grounds that it violates what they see as clean design, and
> at least one of the people objecting to the "install systemd as PID 1
> and then remove it later before ever booting into it" approach seems to
> be doing so on the grounds that that is not a clean design...)
Sorry, I don't do sysvinit vs systemd arguments.
Reply to: