Dear all,
I'm sincerely grateful to technical committee members for their dedication
and relentless effort to thoroughly research and understand the issue in
order to make the best decision possible.
Although most arguments for and against various init systems were already
presented I think I still have something to add. I apologise in advance to
some who might consider my feedback to be obvious or redundant.
This is the first time ever I'm sharing my concerns regarding init system
for Debian.
I think well-balanced decision on this subject would benefit from not being
too technical.
For instance due to controversial contributor's agreement Upstart is pretty
much defunct project. Many contributors prefer to spend their time on
something else rather than Upstart. If adopted Upstart will likely turn into
a big liability for Debian. The very survival of Upstart may depend on
whether we going to be involved or not. Canonical/Ubuntu would be very happy
to use Debian resources for Upstart as if they succeed in "selling" Upstart
to Debian they would be able to offload (i.e. outsource) a significant chunk
of effort that they have to dedicate to Upstart development and maintenance
otherwise. It is quite possible that Ubuntu might reduce their involvement
to Upstart (and "allow" Debian to deal with problems) while they are likely
to spend more of their resources formerly allocated to Upstart to contribute
to other areas of "added value". (IMHO the only major Ubuntu sell point is a
concept of "added value" on top of Debian.) In my opinion Canonical/Ubuntu
will benefit the most from Upstart adoption in Debian.
Considering the possibility that in the future Ubuntu might abandon Upstart,
Debian may end up with unwanted/obsolete init system. Since Upstart future
is uncertain I fear that we might waste a lot of precious resources for
Upstart and/or potentially became de-facto upstream for Upstart. IMHO from
this prospective Upstart shall not be considered as alternative init system
at all.
Indeed I'm concerned about conflict of interests from DDs affiliated with
Canonical and Ubuntu. When they advocate for Upstart I doubt they have
Debian's best interests in mind. There is a danger for Debian to be overrun
by outsiders or to fall under their influence even if some of them are
working on both sides.
Besides we can learn from OpenSUSE where Upstart was replaced with Systemd.
Even without much investigation it should be fairly clear that there are
good reasons not to use Upstart and to prefer something else.
As for Systemd I do not fear its adoption. On the bright side it would be
nice to reduce our differences with other distros in that area. Systemd may
open some exciting opportunities to cooperate and join the efforts with
other influential distros. Our users may benefit from feature rich init
system and its adoption might make it easier for new users to switch to
Debian. It doesn't look like Systemd survival will be influenced much from
Debian involvement so from non-technical prospective Systemd is better for
us due to strong upstream and wide(r) adoption.
Of course there are concerns regarding integration between Systemd and GNOME
but that's a different issue and perhaps not a major one as long as we use
GNOME as default desktop environment. Besides GNOME already became notorious
for being intrusive (e.g. it depends on "pulseaudio" etc.).
Also I'd like to notice that shopping for most feature-rich init system
might be not our goal after all. OpenRC may be the safest choice that might
satisfy majority of developers as it appears to have the least number of
objections. I have impression that OpenRC have far less passionate opponents
than Systemd.
Finally I'm sure everybody is already getting exhausted by long debates
about this topic. At this point it might be tempting to approach on
decision, any decision, to put this to end. This is a way to make mistakes
of judgement. Unless there is a rush we all need to slow down and perhaps
even take a break for several weeks to clear our heads and make a balanced,
well thought decision. Taking break may be beneficial for the quality of
decision making.
--
Cheers,
Dmitry Smirnov
GPG key : 4096R/53968D1B
---
Odious ideas are not entitled to hide from criticism behind the human
shield of their believers' feelings.
-- Richard Stallman
Attachment:
signature.asc
Description: This is a digitally signed message part.