[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: sysvinit is still here, and here to stay for jessie (was Re: systemd is here to stay, get over it now)

That will be my last contribution to this pointless discussion.

Le jeudi, 3 juillet 2014, 16.59:25 Thorsten Glaser a écrit :
> > or without systemd btw). Given that the technical committee has made
> > a decision which stayed unchallenged (so far), I've now come to
> > think that
> No, there just has not been any challenge that met the form and
> other requirements… and I am at a bit of loss at what to do here.
> Besides, it’s not that the TC made a decision. Rather, the TC was
> split, and the chairman threw in his weight.

Sorry, but this is plain wrong (and you know it); the TC followed its 
decision-making process (outlined in the project's constitution [0,1]) 
which led to what is now a TC decision. The detail of the votes doesn't 
change the outcome. Quoting [2]:
> The committee decided that the default init system for Linux 
> architectures in jessie should be systemd.

There are no possible "weak" or "strong" decisions; the outcome of a TC 
or GR decision is either "resolution accepted" or "further discussion". 
Giving importance to the "winning majority" is your choice but doesn't 
change the decision itself.

> This is absolutely not what I’d call a project(!) decision.

I was careful enough to not say it was a project decision, mind you. 
That said, our technical decision body took a decision that stayed 
unchallenged (so far); this fact makes it a /de facto/ project decision.

> > Can we get over this now and start making Jessie the most awesome
> > stable release we've ever prepared together?
> To do that, it MUST work without systemd, if alone for upgrade
> scenarios.

That's wrong: as far as the default init is concerned, it only MUST work 
without systemd-as-init until the first post-dist-upgrade reboot. I 
don't think any work outside of that scope should be imposed on the 
package maintainers [3].

> And alone the fact that the systemd issue *continuously* pops up
> shows you that it is nowhere even near solved.

I don't think that's true. From my (most-certainly) biased point of 
view, the issue continues to pop up because some very vocal systemd 
opponents keep vocally insisting that other fellow project volunteers 
ensure that their pet use-cases keep working with a non-default init 
system (or even without a specific non-init package). As far as I am 
concerned, I'm putting my energy to make sure my packages (and my setup) 
work as good as possible with the _default_ init that was picked for 
jessie. The rest is best-effort.

> Furthermore, the TC(-chairman) decision only was on the default
> init system for the Linux ports of jessie.

Please stop spreading the FUD that the default init decision was a TC-
chairman decision. This is plain wrong, as our constitution outlines.

> This means that
> • installing jessie with other init systems
> • switching between init systems
> • default init system for kFreeBSD ports
> • default init system for Hurd port
> • which non-default init systems are there?
> are still on the table.

Sure. They are on the "we want Debian Jessie to work with other inits 
than the default" persons' table, they have no reason to be on 
everyone's. As mentioned above: if you want this to work, make sure it 
does, propose patches, test scenarios, etc. Pushing for a package 
conflicting with systemd is not helping any of this.

> (Due to Debian’s requirements for sane upgrades, running a jessie
> system that was upgraded from an older release with sysvinit MUST be
> fully supported, anyway.)

Wrong, see above.

> I’m a bit torn between throwing it all (which is a bad idea ofc),
> writing a GR myself (which is also a bad idea due to my lack of
> language skills), packaging BSD init for my own repo, joining the
> (currently unheard) runit-as-init crowd…

Frankly, if voting on a default init GR would ensure we'd finally stop 
these bikeshed discussions (after few other weeks of mudslinging), by 
all means, go for it. That said, as far as I remember, the latest GR 
proposal [4] on this subject failed to gather the mandatory K seconds 
though. For me, this indicates that not even K=5 DDs were interested in 
having that discussion. We currently need half a percent of DDs to 
trigger a GR and it has not happened; could we get over this now?


[0] https://www.debian.org/devel/constitution
[1] Which you agreed to.
[2] https://www.debian.org/devel/tech-ctte
[3] It doesn't mean they should not accept patches for adaptations for
    other init systems though.
[4] https://lists.debian.org/debian-vote/2014/03/msg00000.html

Reply to: