I'm happy to see that Debian has reached a decision about its default init system, and that the Ubuntu team have been gracious about it and opted to support their downstream.

What will probably surprise most people is that I think it's the right decision. systemd has the momentum out of the two projects, and has both the technical and community leadership advantages.

Friendly competition is a brilliant development model until there is a clear winner, at which point friendly collaboration tends to work out better. It's time for that transition.

I hadn't posted anything of this effect until now because I didn't want my position to influence Debian's decision ("Original Upstart Author endorses systemd!"). That wouldn't have been fair on the current Upstart maintainers +James Hunt+Dimitri John Ledkov and +Steve Langasek who have been doing an absolutely fantastic job continuing the project I started.
159
25
Eugene Sliusar's profile photoEdvin Alvarado's profile photoPaul Covell's profile photoPatrick Plattes's profile photo
27 comments
 
Which leaves the question: what will Google do for Chrome OS?
 
+Andreas Proschofsky: probably nothing in the short term, re-engineering our boot sequence around systemd is probably a lot of work and it's unlikely any engineer is going to want to do that just for the sake of it.

We haven't actually updated Upstart in a few years anyway, we can carry on just fine with it indefinitely without any downside.

If someone stepped up, did all the hard work, and demonstrated no regression in boot time on actual Chrome OS hardware, nobody would block switching either.
 
Which leaves the question: could upstart ever get the momentum and community leadership advantage, or was it prevented from succeeding on technical merit by the sacred Canonical CLA+GPL combo ?
 
+Thierry Carrez: could it have? (past tense). Yes. I think the CLA was what held it back.

Can it now? I don't think so. There are more interesting things for people to work on.
 
So, say someone were to re-engineer the ChromeOS boot sequence around systemd, would they have to sign the Google CLA? :)
 
Thanks Scott, I'm delighted to see that chances are good that this shift is going to be a positive thing for everyone... Posts like this really help.
 
I'll tell you what I told Lennart: I never thought there was anything wrong with init. :)
 
+Thierry Carrez umm... I had to sign OpenStack CLA and even join Foundation to contribute to OpenStack. So sacred CLA didn't stop OpenStack. Why would it stop Upstart? 😉
 
Didn't the CLA change language over the course of Upstarts life?
 
+Ante Karamatic the test will be if there's a competing free product with no cla. Not sure that's true for OS yet. From memory Eucalyptus is also cla'd. 
 
+Martin Pool OK, how about Apache? Apache's CLA covers not only code but ideas too. Yet, Apache is the most popular web server in the world. Cause it's awesome.

Let's be honest and say that whole story about CLA was basically FUD. I would agree that Canonical CLA pre-2011 was suboptimal/bad, but that one is not in effect for last 3 years (+Tim Chavez).

There's nothing wrong with CLAs, they actually protect the project in cases where licenses fail to do that.
 
As a previous lead of a project under both Canonical CLAs and as someone who was involved in launching their current CLA, I do believe it discourages a significant number of contributors. Not necessarily enough to kill a project but it's a competitive disadvantage.

It is an expensive choice to require a strong CLA, and often not a good deal for the supposed beneficiary.

The Apache weak CLA is a different beast. 
 
+Ante Karamatic The OpenStack CLA is not a copyright assignment, and it is used with a permissive license. Canonical's Copyright assignment + GPL is the combination what makes other contributors afraid to participate, because it gives total control to a central entity.
 
Yeah, but you can sign new CLA and work on it today. You could've done that for 3 years.
 
+Ante Karamatic I'd argue that the original CLA+GPL setup discouraged people to join in 2009, and them invest in an alternative in 2010/2011. Once they started, they grew attached to it... Also I know several companies that don't like the sublicense provisions in the new format, so it's still a blocker to some (which are fine with the weak, patent-grant based OpenStack CLA).
 
Don't get me wrong: I think the OpenStack CLA should also die a horrible death. Weak as it is, it's still a useless barrier to contribution.
 
+Ante Karamatic the current CLA only marginally helps. With GPL plus unlimited license to a commercial entity clearly the community is resistant to contributing.
 
I wonder what the resistance is actually rooted in?
 
+Tim Chavez most open source contributors do so from a place of at least some principle. Allowing a single company to secretly violate those principles using code you have written is just unappetizing. The maintenance burden isn't enough to override that instinct in most of these individuals. Combine that with corporate red tape and you get almost no contributions of significant effort.
 
Thanks Scott - very gracious attitude all round - and I'm sorry I put you on the spot earlier. Well, a little bit ;)
 
+Scott James Remnant Very gracious post dude.  Even without direct lines of code upstart has/will continue to influence general design trends and the whole ecosystem, so for that I thank you.
Add a comment...