On 11/29/2014 08:34 AM, Florian Lohoff wrote:
> Hi,
>
> On Fri, Nov 28, 2014 at 08:58:19PM +0000, chris wrote:
>> Aside from fixing Debian(!) and other than being anti - systemd, it
>> seems unclear to me what the aims and goals are for the project.
> For me its not anti-systemd - but pro-choice.
>
> I want to have the ability to install a Debian without touching
> systemd. You cant right now. You can choose to run sysvinit as
> pid1 but only by purging systemd and reinstalling sysvinit. But still
> then you are left with a whole lot of the systemd ecosystem.
>
> Have a look at the rdepends of libsystemd0 and other elements
> of the systemd ecosystem.
Right.

Or more generically: We must base our goals on principles rather than
particular software packages we disagree with.

We should not hate systemd, or Lennart, but the way it has been handled
in hijacking the system we knew and loved. And I think we can learn some
things from systemd too, but we would have to create something
consistent with the unix philosophy and common sense before we can use
it, not because it is impossible to do it otherwise, but because it
would be inefficient and dangerous.

And FYI about me and my opinion... I have been using openSUSE on
desktops and home servers for a decades, and Ubuntu LTS on servers at
the office (which was their choice before I joined them), and since
systemd has been integrated into openSUSE, it has more strange bugs than
ever... raid being degraded on boot by removing random disks, systemd
processes eating 100% cpu for no reason and have to be killed, lots of
useless error messages replacing useful ones in syslog (eg. when a
daemon won't start but is set to auto start after failure, it is hard to
find the real cause and you only find "systemd[1]: xxxx.service start
request repeated too quickly, refusing to start."... you can't just run
"restart" again to see a new log with actual problem, since it will just
refuse, and can't even test your fix until this arbitrary timer is reset
wherever it is controlled.), and on Ubuntu 14.04 with some sort of
systemd-udev but not all of systemd, you randomly get network interfaces
named "renamed3" and so on, because of bugs, for a useless feature I
don't want (only useful for machines without stable MAC and rw root fs).
These things come from it being beta and forced on us anyway... surely
they could be fixed, but the development process is all wrong, so it
makes no sense to work with such a large mess. We should stick to
processes that work, and that people actually agree with without huge
corporations forcing them to (beta test it for them?). If we want to
beta test something, we will do it out of our own free will.

And I'd like to contribute, but not sure how... I could ask about
putting some services at the office, but we don't have decent bandwidth
(we have 100 Mbps uncapped but we already push it to the limit). But I'm
fully in charge of the IT there. And it probably isn't an issue to run a
non-HA VM there. And my reason for joining here is because I want to
promote freedom of choice, and I don't see another binary distribution
doing that. And it's not just systemd... freedom is part of the
philosophy; we can't abandon that goal for any reason other than gaining
more freedom than you sacrifice (such as with GPL copyleft--freedom to
contribute without the fear that your work will be used against you).

>
>> Which brings up in my mind another issue, who's in charge? is in a
>> dictatorship? is there any kind of formal issue resolution ?
> Without beeing associated more than you i guess this question comes
> to early.
>

This is not a simple topic.

Linus Torvalds is a benevolent guy, and does an exceptionally good job,
but I would hesitate to trust a system that can be so easily hijacked by
the next falsely benevolent usurper. However, a free software license
based dictatorship is not nearly as dangerous since you can just fork it
when necessary (but the corrupt fork still can steal the name,
infrastructure, donations, etc. along with future contributions
associated with those things).

And plain traditional democracy does not solve anything... it just gives
you an excuse for the majority to violate the minority, and the illusion
of public support... which is worse than a dictatorship because you
don't know who controls it, and there is no control on what the
representatives can do (and there can be no control... it has to be a
whitelist rather than a blacklist to have any chance), so a well funded
group can easily corrupt it. (And well funded members are far more
active, so they don't even need a majority).

I don't know specifically what to say here except that I hope we don't
choose something blindly democratic that is flawed and lets us fail like
Debian has. We need to learn from the mistakes of the past, and try to
prevent them, at least by noticing them early and being reactive, but
ideally proactive.