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

Re: Bug#727708: Fsck SystemD and its developers and its users. GR to override this please.

On 02/11/2014 04:31 PM, Clint Byrum wrote:
> One point of moving to a system like upstart or systemd is that the
> sysvinit scripts do not run as scripts. They are little tiny declarative
> files that run all or most in C. This speeds up boot, but only makes
> sense if all of the early stage boot things make use of it.
> Leaving most things to just use the sysvinit compatibility layer means
> not realizing one of the more important benefits of the default init
> system if it should in fact turn out to be systemd.

I agree with the above.

> So at best you're talking about maintaining two for every daemon. That
> is still roughly twice the maintenance work and twice the testing.

Yes, that's my proposal, and as well deprecate sysv-rc in the favor of
OpenRC, and allow OpenRC runscript files *only* if there's support for
the default init system (because this way the default init system will
not use them, so the runscript format is possible). Yes, testing and
maintenance will be double the amount of work, which is why I wrote that
it will all depend on contributions. I didn't write support for it would
be mandatory, but on best effort basis, which I think is fair. Also,
that will enable support for our non-linux ports.

> Not saying I like it, but that is where choice hurts Debian. Perhaps
> having the choice will also help Debian enough to make it worthwhile.

That's what I believe, yes.

On 02/11/2014 05:03 PM, Andrei POPESCU wrote:
> According to Russ Allbery, it's easier to maintain both systemd and
> upstart declarations than one sysv init script.

s/sysv init/sysv-rc/ <--- Please don't do this mistake, it's really
sysv-rc that uses the init scripts, and OpenRC, which has a declarative
format, still uses sysvinit.

Anyway, to some degree, I agree with Russ here, which is why I think we
should replace sysv-rc by OpenRC completely at some point (which is when
we have stabilized support for *all* arch, which isn't the case right
now, latest Hurd patches broke support for kFreeBSD).

On 02/11/2014 05:31 PM, John Paul Adrian Glaubitz wrote:
> On 02/11/2014 09:02 AM, Thomas Goirand wrote:
>> The situation with upstart or systemd, if not chosen as default,
>> will be quite different, since not all packages are supporting them
>> directly right now. One of these 2 will suffer from the choice of
>> default init system.
> What? That's not true. As people have explained here before - even
> directly to you - both Upstart and systemd have perfect backwards
> compatibility with sysvinit scripts.

You missed one very important word: *directly*. Probably I should have
write *natively*. Sorry for this.

On 02/11/2014 05:27 PM, John Paul Adrian Glaubitz wrote:
> No, it's absolutely not. You can have the choice for the interior
> design, the paint job, the radio, the type of engine and comfort
> features, but you certainly cannot have the choice on internal
> parts like the ignition system or starter motor.
> Furthermore, if you do decide to replace these parts on your own,
> you will end up losing your car manufacturer's warranty.

The car analogy can only go so far... My point was saying that people do
customize things, and do it. That's a way more the case with computers
than with cars.

On 02/11/2014 05:27 PM, John Paul Adrian Glaubitz wrote:
> And since there are virtually no volunteers for OpenRC besides
> you and the other two OpenRC maintainers, Roger and Benda, it
> will be unsupported at some point when you guys step down.

That's truth for every bit of Debian, however, package gets orphaned,
adopted, etc. Please don't through this type of argument, especially
when we have co-maintainers already.

On 02/11/2014 05:27 PM, John Paul Adrian Glaubitz wrote:
> I have seen you asking for help on OpenRC so many times during
> these discussion, but I am yet to see people raise their hands
> and say "Yes Thomas, I am going to help you!"

Yet, it happened. Not in this list though...

On 02/11/2014 05:27 PM, John Paul Adrian Glaubitz wrote:
> All I read are statements from you like "Yes, it would work
> in general if we had someone to implement it, I don't have
> the time right now unfortunately."

Please read the debian/changelog of OpenRC in Experimental, and measure
the work that has been done so far. I don't think this counts for
nothing, and I really see progress. Hurd & kFreeBSD support, and
lsb2rcconf comes to mind. Plus this doesn't include all what has been
done before the package entered Debian.

> It's not *my* choice, systemd is the choice of the majority of the
> Linux community. OpenRC and upstart are used in Gentoo and Ubuntu
> only (ChromeOS doesn't really count in that context, it's a more
> or less closed system by Google), while virtually every other
> of the large distributions has adopted systemd.
> Using something which is not widely adopted and has very few
> supporters in the development community means that if any of the
> OpenRC or Upstart people will decide to retire, these systems will
> lose much more development manpower than systemd does.

Please stop spreading useless FUD. I could do that saying: "what happens
if both Keith and Lennart stop developing systemd", but you would
rightly find it very silly, no?

Plus I'm replaceable, if others see enough interest in the package
(which seems to be the case). If I'm not mistaking, there's already 6
committers in the Git on Alioth. Hacking an init system is a lot of fun,
and I think that's part of the reason why there's contributors. It is my
hope that the OpenRC community continues its growth.

Thomas Goirand (zigo)

Reply to: