systemd for 11.10 ?

[Clint Byrum]

Excerpts from Phillip Susi's message of Sun May 08 15:11:57 -0700 2011:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 05/06/2011 07:03 PM, Oliver Grawert wrote:
> > hi,
> > Am Samstag, den 30.04.2011, 18:24 +0200 schrieb Thomas Bechtold:
> >> Hi,
> >>
> >> i just want to know if there are any plans to replace upstart[1] with
> >> systemd[2] for 11.10?
> 
> I have to say I was skeptical at first, but after reading that very good
> introduction it sounds great!
> 
> > http://undacuvabrutha.wordpress.com/2011/04/29/why-ubuntu-should-continue-with-upstart-for-11-10/
> 
> Seriously?  It is too close to LTS?  It's over 11 months away!  I don't
> mean to jump on the unity haters bandwagon, but that was pressed in one
> dev cycle and was still in pretty bad shape just 2 months before
> release.  Surly if we can manage that, we can manage systemd in 11
> months?  Or at least make an attempt; you can always switch back if it
> is still looking bad at beta time.

Hi Phillip, I appreciate that it seems like a change similar to the switch
of default desktop environment, however, there's a massive difference.

Changing the UI meant a great deal of progress toward the project's stated
goals. Changing the pid 1 just means adding efficiency to something that
already works, and is starting to be well understood. Its Diesel versus
Gasoline. Each has their merits, but the overlap is 80% and the 20%
difference is, IMO, mostly corner cases and optimizations.

Change has a cost, and no matter how much of that cost we pay up front
in man power, we simply cannot get it done in time without taking on
technical debt, because some of the debt is the unknown bugs and corner
cases that a change brings. Since LTS also means incurring a greater
maintenance cost due to its extended life cycle, this ultimately means
LTS multiplies some of the technical debt. So, we should strive for
extremely low technical debt the closer we get to an LTS, so that we're
not paying it back as we fix bugs for the lifecycle of the project.

9.10 added some of that debt by putting in place upstart before it
was widely understood, and before it had been sufficiently utilized to
implement complex dependencies (see bug #525154 for an example). That
debt has since been paid in user frustration and painful debugging.
The investment, however, paid off in faster boot times, so the trade
off wasn't all bad.

We're getting close to paying most of that off now. To consider taking
on even more would have been a bid mad IMO. The things it adds *are*
interesting. However, IMO they are all optimizations and features to
help with corner cases we've already covered, or things that make the
parallelized, different style boot easier to swallow. Our users have
already swallowed event based boot, as bitter a pill as it was for some
of them. Lets thank them by providing them a smoother, more refined boot
experience for 11.10 and 12.04, then consider systemd once we're ready
for some hard core refactoring in 12.10.